AGENT00

Dashboard

Loading…

Open To-Dos

–

Today's Events

–

H100 Hrs (MTD)

–

Backup

–

files

Memory

–

entries

DB Size

–

Git Push

–

Last Load

–

System

Soul ●

Agents ●

API Usage

$–

–

Version

–

build

■ To-Do

▦ Agenda — Upcoming

Loading…

◷ H100 Advisory

◈ Memory

Ask Anything

Conversation with soul prompt + memory · full context retained

Start a conversation — ask about markets, strategy, your portfolio, or anything else.

Trading Intelligence

–

RED FOLDER

Loading…

SIGNALS

Loading…

EARNINGS

Loading…

▣FUTURES

Loading markets…

▣MACRO

Loading macro…

Loading economic calendar…

📡NEWS — LIVE HEADLINES

Loading headlines…

💵USA COMPANY EARNINGS

—

Loading earnings calendar…

Loading yields…

●US TREASURY RATES & YIELD CURVE

Loading rates…

Loading yield curve…

●RATES REGIME FRAMEWORK

Regime	2Y Range	10Y Range	2–10 Spread	30Y Mtg	Macro Context	Equity Risk	Housing / Credit
R0 EASY LIQ	0–2%	1–3%	+50 to +150bp	3–5%	Post-recession / QE: easy policy, low term premium	Very bullish for beta, growth, high duration	Strong housing, easy refi and credit conditions
R1 GOLDILOCKS	2–3.5%	3–4.25%	0 to +100bp	5–6.5%	Near-neutral policy, inflation anchored	Constructive if earnings growing; valuation-driven	Housing slows from boom but still functioning
R2 HIGH TOLERABLE	3.5–4.5%	4–4.75%	−25 to +75bp	6.5–7.25%	Higher for longer: policy restrictive, term premium rebuilding	Pressure on high-multiple growth; value outperforms	Affordability stressed; volumes fall but system intact
R3 BEARISH RATES	4.5–5.5%	4.75–5.5%	−50 to +75bp	7–8%	Real yields compete with equity earnings yield; cash is alternative	Structurally bearish; high-duration tech hit hardest	Housing recession risk high; refi window shut
R4 DEEP INVERSION	3–6%	2y > 10y	−50 to −150bp	Follows 10y	Pre-recession: tight current policy, cuts priced out the curve	Medium-term bearish; recession risk high	Housing weakens on job/income risk; credit tightens
R5 EASING / RECESSION	0–3%	1.5–4%	+100 to +250bp	Falls but lags	Fed cutting aggressively into/after recession	Bear market / early recovery: add risk selectively	Housing stabilises slowly as rates fall
R6 TERM PREMIUM SHOCK	2–4%	4.5–6%	+75 to +200bp	7.5–9%	Long end sells off on term premium / fiscal / inflation fears	Very bearish for duration and credit; steep for wrong reason	Severe affordability shock; volumes collapse
R7 SUPER TIGHT	6%+	7%+	Flat / inv / steep	9%+	Extreme inflation fight: ultra-tight policy, very high real yields	Brutal for risk assets until inflation breaks	Severe pressure; delinquency and default risk rises

Current regime auto-detected from live 2Y/10Y/30Y yields and 2-10 spread. 30Y ≥ 5% for 10+ of last 20 days escalates to high-alert sub-regime.

●10Y UST YIELD REGIME

Regime	10Y Range	Real Yield	Macro / Policy	Equity Implications	Notes / Criticality
T0 EASY <3%	<3%	Real ≤0–1%	Very easy / QE-like or post-recession; growth fears or disinflation	Strong tailwind for long-duration growth; high multiples sustainable, ERP wide	Only likely in serious slowdown or renewed QE; bullish but coexists with bad macro
T1 FRIENDLY 3–4%	3.0–4.0%	~1–2% real	Goldilocks: inflation anchored, policy near neutral, modest term premium	Equities comfortable: multiples supported if earnings grow; growth vs value idiosyncratic	Equity-friendly band. Crossing back above 4% often marks end of easy conditions
T2 PRESSURE 4–4.75%	4.0–4.75%	~2–2.75% real	Restrictive but not yet breaking: higher-for-longer, term premium rebuilding	Valuation headwind: high-duration growth de-rates; quality/value outperform	4% is key transition level. Above ~4.5% triggers systematic de-risking
T3 COMPETITIVE 4.75–5%	4.75–5.0%	≥2.5–3%+ real	Bonds compete with equities: real yields clearly tight, fiscal concerns front & centre	10Y competes with S&P earnings yield; ERP compresses, broad multiple pressure	Critical band: sustained high-4s treated as bearish unless earnings exceptional
T4 RISK BREAKS >5%	>5.0%	>3% real	Persistent inflation, serious term premium / fiscal risk, or policy error	Historically → equity drawdowns and cross-asset stress; sharp re-rating of duration	Sustained >5% triggers rates-breaking-things regime, especially with 30Y/mortgage elevated

10Y regime auto-detected from live yield. Real yield interpretation assumes ~2–2.5% inflation baseline.

Regime	10Y Range	Real Yield	Macro / Policy	Equity / Housing Implications
G0 EASY <3%	<3%	Real ≤0%	Post-crisis / QE: BoE ultra-accommodative, gilt scarcity	FTSE growth rally; housing booms, mortgage rates very low
G1 FRIENDLY 3–4%	3.0–4.0%	~0.5–1.5%	BoE near neutral, inflation anchored ~2%, term premium modest	FTSE comfortable; housing stable, mortgage rates manageable (3–4%)
G2 PRESSURE 4–4.75%	4.0–4.75%	~1.5–2.5%	BoE restrictive, inflation persistent, fiscal risks from gilt supply	UK equities face valuation headwind; housing slows, SVR mortgages 6%+
G3 STRESS 4.75–5.5%	4.75–5.5%	~2.5–3%+	Gilt market stress: fiscal credibility concerns, mini-budget territory	Sterling pressure, pension fund LDI risk, housing affordability crisis
G4 CRISIS >5.5%	>5.5%	>3%	Gilt crisis: loss of fiscal credibility, forced BoE intervention	Broad UK asset sell-off, sterling collapse, housing market freeze

▣REGIME STRIP

Cross-asset day desk · operating environment

Warming up regime cache — first refresh in ~10 seconds

▣DRIVER MATRIX

What is pushing each focus market right now

Warming up driver matrix — first refresh in ~15 seconds

▣FOCUS WATCHLIST

Execution-first intraday metrics

Loading watchlist…

▣RISK RAIL

Warming up risk rail — first refresh in ~15 seconds

▣CAUSE & EFFECT

Lead-lag and impulse response

Loading cause & effect…

Loading…

▣NEWS REACTION TAPE

Filtered to high-signal events and measurable market response

Impact ≥ medium

Loading news reactions…

▼ Legacy Market Technicals

▶ICT MULTI-ASSET BRIEF

Click ▶ Regenerate to generate fresh analysis across all assets.

■EOD SUMMARY

Click ▶ Run EOD to generate end-of-day summary, or ↻ to load last cached.

Strategy →

Click refresh to load performance stats.

📊 Strategy Health — Expectancy Engine

Loading expectancy data…

📡 TradingView Alerts

TradingView Webhook Setup

1. In TradingView, create an alert on any chart
2. Under Notifications, enable Webhook URL
3. Paste your webhook URL: https://honeybees-mac-mini.tail56e8bc.ts.net/api/tv/webhook
4. Set the Message to JSON format:

{"ticker":"{ {ticker} }","action":"{ {strategy.order.action} }","price":"{ {close} }","interval":"{ {interval} }","message":"{ {strategy.order.comment} }"}

5. For simple alerts use: {"ticker":"{ {ticker} }","action":"alert","price":"{ {close} }","interval":"{ {interval} }","message":"Your alert name"}

No alerts yet — set up webhook in TradingView to receive real-time signals.

⚡ ATR Context — Daily Range Used vs Remaining

Loading…

📏 Range Monitor — Today vs 20-Day Average

Loading…

🎯 Signal Quality Score vs Outcome

Loading…

🔥 Session P&L Heatmap — Win Rate by Asset × Kill Zone

Loading…

📉 Drawdown Monitor — Session P&L & Risk Status

Loading…

✅ Confluence Breakdown — Factor Analysis per Signal

Loading…

⚡ Kill Zone Performance — Win Rate & Expectancy by Asset

Loading…

💧 Liquidity Profile — Typical Hourly Volume by Asset

Loading…

AMD 50% Range Rebalance — Power of Three

Strategy: the strategy author's 50% rebalance model. Price always redelivers to 50% of the dealing range (EQ). Wait for Daily PO3 manipulation wick above/below prev day H/L → H4 PO3 at 10am EST → SMT divergence NQ/ES → Inversion FVG entry. Target = 50% EQ. BE = when H1 candle flips. Primary asset: MNQ. Also works on BTC (daily timeframe).

Click Scan Now to run the AMD 50% scanner.

AMD 50% Strategy Rules

SHORT Setup Checklist

🔴 Price in PREMIUM — above 50% EQ of dealing range
🔴 Daily candle wicks ABOVE previous day high (manipulation)
🔴 H4 sweep of H4 range high at or near 10am EST
🔴 SMT: NQ makes new high, ES fails to confirm
🔴 Inversion FVG: previous bullish gap breaks, retap as resistance
🎯 Target: 50% EQ of current H4 dealing range (base hit)
🛡 Stop: above the manipulation wick high
⚡ BE: immediately when H1 candle flips bearish

LONG Setup Checklist

🟢 Price in DISCOUNT — below 50% EQ of dealing range
🟢 Daily candle wicks BELOW previous day low (manipulation)
🟢 H4 sweep of H4 range low at or near 10am EST
🟢 SMT: NQ makes new low, ES fails to confirm
🟢 Inversion FVG: previous bearish gap breaks, retap as support
🎯 Target: 50% EQ of current H4 dealing range (base hit)
🛡 Stop: below the manipulation wick low
⚡ BE: immediately when H1 candle flips bullish

AMD 50% Edge Rules: "You either need to be right or right out." · Base hit every day (50% → 50%) — don't try to catch the extreme. · Break even aggressively — in his words this is what makes the model profitable. · Session window: 9:15–11:30 EST (9:30 liquidity injection + 10am manipulation reversal). · Do NOT trade around the 9:30 open unless confirmation is extremely clear. · 4-week data gate: track this strategy in AGENT00 and patterns will emerge after 28+ trades.

LIQ+VOL — Intraday Liquidity & Volatility

Strategy (Jadecap): Markets raid resting orders at previous session highs/lows before reversing. Wait for the liquidity sweep of Asian/London range or prev-day H/L → zero follow-through → closure back through the level → entry. SMT divergence (NQ≠ES) confirms. Target = opposite session extreme. Session window: 9:30–10:30 EST (70% of intraday H/L printed here). Min R:R 1.5:1 · Midnight open as daily reference. signal Saved to TRADING:LIQVOL:LATEST

Click Scan Now to run the LIQ+VOL scanner.

LIQ+VOL Rules

Bias & Setup

Mark previous 1-3 day highs/lows (resting stops)
Mark Asian range (8pm–midnight EST) + London range
Wait for raid of one level — do NOT anticipate
Zero follow-through after the sweep = signal
Closure back through the raided level = entry trigger
SMT: NQ sweeps high, ES fails (or vice versa) = confirmation
FVG / market structure shift on H1/15M for precision entry

Execution & Management

Window: 9:30–10:30 EST primary (70% of intraday H/L)
Target: opposite session extreme (Asian low → target Asian high)
If both sides raided → expect consolidation, stand aside
Exit: time-based — before 11am/noon if not at target
Scale-in: only after reducing risk on original position
Min R:R: 1.5:1. Win rate ~35% — management is the edge
4-week gate: track all signals before drawing conclusions

Key insight: Win rate ~35% — profitability is entirely from trade management. Average win must be 2.5–3× average loss. Selectivity and execution > entry accuracy. "You either need to be right or right out." — Do NOT widen stops when price approaches SL.

Mean Reversion — Capitulation / Right Side of the V

Strategy (Lance Breitstein — $100M+ verified): Stable assets making unusually fast/large moves on no fundamental news are out of equilibrium. Wait for the turn (right side of the V). Score variables 0-10 each. Higher total score = larger position. Target = 20-period MA (Bollinger midline). Works on all assets and timeframes. Fractal by nature. signal Saved to TRADING:MEANREV:LATEST

Click Scan Now to run the Mean Reversion scanner.

Mean Reversion Variable Scoring

Size & Speed (0-20)

Size of move — vs average daily range. 3×ATR = high score
Rate of change — asymptotic waterfall = maximum score

Context (0-30)

No fundamental news — clean move = high score
Days in a row — 5+ consecutive bars same direction
Forced liquidation — volume spike = capitulation signal

Asset Quality (0-30)

Boringness — stable historical range vs current move
Market cap / diversification — S&P > stocks > crypto
Quantifiability — bonds > stocks > BTC > micro cap

Entry (Right Side of the V)

NEVER buy front side — wait for the turn
Entry signals: break of downtrend line · prior bar high break · lower-low/lower-high pattern breaks
Extreme only: intra-bar turn acceptable if move is truly asymptotic (tiny size)
Stop: lows of the move (long) / highs of the move (short)
Target: 20-period MA (Bollinger midline = equilibrium)
Trail: prior bar lows (long) / prior bar highs (short)

Score → Position Sizing

Score 80-100: A+ — maximum size, full conviction
Score 60-79: B — standard size
Score 40-59: C — small size only
Score <40: No trade
Win rate: ~70-80% on A+ · ~50% on C
Scale-in: add only after first position at break-even or better

Lance's key insight: "The biggest mistake is not thinking about all variables — people oversimplify. If X then Y never works. It's all variables on a spectrum." Selectivity is everything — trade the 90/100 charts heavily, ignore the 50/100 charts completely.

Opening Range Breakout — 5m + 15m

Strategy: Mark the high/low of the first 5 or 15 minutes after the NY cash open (09:30 EST). Trade breakouts of that range with HTF bias + volume confirmation. Targets 1× OR range. Stop on the opposite side. Primary assets: MNQ, MES. Also scans MGC and MCL.

Click Scan Now to run the ORB scanner.

ORB Strategy Rules

LONG Setup Checklist

🟢 OR marked 09:30–09:35 EST (5m) or 09:30–09:45 EST (15m)
🟢 Price breaks above OR high
🟢 Volume on breakout bar ≥ 1.5× average
🟢 RSI14 (1m) between 55–75 — momentum building, not exhausted
🟢 1H bias BULLISH — price above 21 EMA
🟢 OR width within clean range (0.05%–1.0% of price)
🎯 Target: OR high + 1× OR range
🛡 Stop: OR low
⚡ BE: trail to entry when 1× OR reached

SHORT Setup Checklist

🔴 OR marked 09:30–09:35 EST (5m) or 09:30–09:45 EST (15m)
🔴 Price breaks below OR low
🔴 Volume on breakout bar ≥ 1.5× average
🔴 RSI14 (1m) between 25–45 — momentum building, not exhausted
🔴 1H bias BEARISH — price below 21 EMA
🔴 OR width within clean range (0.05%–1.0% of price)
🎯 Target: OR low − 1× OR range
🛡 Stop: OR high
⚡ BE: trail to entry when 1× OR reached

ORB Edge Rules — Momentum Strategy: ORB is a pure momentum play — the edge is breakout impulse, not R:R width. Minimum R:R = 1.0× (not 1.5×). · RSI14 (1m) is the key confirmation — 55–75 for longs, 25–45 for shorts. RSI >80 or <20 = exhausted, skip. · 5m variant is faster but noisier — tagged as ORB_5M. · 15m variant is slower but cleaner — tagged as ORB_15M. · Session window: 09:30–11:30 EST — do not trade after 11:30. · Score ≥7/10 auto-creates a monitored signal with Telegram alert. Each variant gets its own 1-active-per-asset slot, so ORB_5M and ORB_15M can both be active on MNQ simultaneously.

ICT Setups — Today

■ All Execution History

Loading history…

Signal Monitor — Live Indicators by Asset & Timeframe

Click refresh to load…

Asset Analysis

LLM · live data

Select an asset above to run analysis…

Key S/R Levels — All Assets

Click refresh to load key levels…

Market Data

Global Markets · Indices · FX · Commodities

Markets, indices, ETFs, FX, commodities, rates, correlation and equities have been moved to
Trading Intelligence → Overview sub-pills.

Loading…

📰 H100 Headlines

Loading headlines…

Quick Snapshot →

Select an asset above

Bitcoin Intelligence

AGENT03 · H100.ST · Peer Analysis · Research

BTC Spot

—

ATM IV 7d

—

ATM IV 30d

—

ATM IV 90d

—

RV 30d

—

VRP 30d

—

IV − RV

IV/RV Ratio

—

Regime

—

RV 7d —

RV 60d —

25Δ RR 30d —

25Δ BF 30d —

Venue Deribit

OI Share ~85%

■ Market Context & Regime

Loading analysis…

■ Implied Volatility Surface

Low

High

Strike × Tenor · Ann. %

The IV surface plots implied volatility across strikes (moneyness %) and tenors. Warmer colours = higher IV. OTM puts (left) typically trade richer than OTM calls (right) — this is the skew. Hover any cell for exact IV.

■ 25-Delta Risk Reversal & Butterfly

Loading…

■ 10-Delta Risk Reversal & Butterfly

Loading…

■ ATM Term Structure

■ Top Ranked Strategies — Live at Spot

Loading…

■ Exchange Landscape

Deribit ~85% global OI · institutional flow, block trades, IV price discovery · WebSocket real-time · CME ~8% · regulated US benchmark · Binance, OKX, Bybit retail/semi-institutional · BTC IV persistently 60–100%+ ann. vs equities 15–25% → structural VRP.

■ Full Surface Explorer

■ Smile Slices by Expiry

Each line shows the IV smile for a single expiry across moneyness. A steep left wing (OTM puts) means downside protection is expensive — typical in BTC due to structural demand from funds and treasuries. If the smile flattens at longer tenors, the market expects mean-reversion. If short-dated smiles are steeper than long-dated, near-term tail risk is being priced aggressively — consider whether that premium is worth selling or whether protection is warranted.

■ Forward Volatility Ladder

Loading…

Forward vol is the market’s implied expectation of volatility between two future dates, extracted from the term structure. When forward vol is significantly higher than spot IV at the same tenor, the market expects a vol increase — potential catalysts like halvings, ETF decisions, or macro events. When forward vol is lower, the market expects vol compression. Calendar spreads exploit dislocations between spot and forward vol. A positive Δ (forward > spot) favours selling the back leg; negative Δ favours buying it.

■ OI Concentration & Strike Crowding

Loading…

■ Gamma Exposure (GEX) by Strike

Loading…

■ Strategy Screener — Ranked by Composite Score

—

Loading strategies…

■ Scoring Framework

Composite = 0.30(Relative Value) + 0.20(Carry) + 0.20(Treasury Fit) + 0.15(Liquidity) + 0.15(Stress Robustness) · Range 0–100 · Mandate-weighted variants for Yield/Accumulation/Protection/Tactical RV

BTC Holdings

BTC NAV

—

Net Premium Received

Options P&L

0% on prem

Eff. Sale / Buy

—

Covered Calls

0 BTC

VaR 1d 95% (BTC only)

—

VaR 1d 95% (Net)

—

Options VaR Impact

—

Max 1d Loss (99%)

—

NET DELTA (Holdings+Opts)

0 BTC · $0

OPTIONS DELTA (BTC)

GAMMA P&L per 1% move

0% NAV

VEGA P&L per 1% vol

0% NAV

THETA Income per day

$0/d

0% NAV/d

■ Add Position

TYPE

DIRECTION

STRATEGY

STRIKE ($)

MONEYNESS → STRIKE

—

EXPIRY

TRADE DATE

QUANTITY (BTC)

PREMIUM ($ per BTC)

PREMIUM %

—

■ Governance

Max overwrite: 15–25%

Max CSP: 10–20%

Naked short: Prohibited

■ Active Positions — P&L

No positions — add one to start modelling

■ Portfolio Payoff at Expiry

Total portfolio P&L (BTC holdings + options) at various spot prices at expiry. Green = profit, Red = loss.

■ Portfolio Greeks (1st + 2nd Order)

Loading…

■ Greek Stress

Loading…

■ VaR & Expected Shortfall

Loading…

■ Historical VaR (1d 95%)

Rolling 30d VaR based on BTC holdings × realised vol. Shows how portfolio risk evolves over time.

■ Theta P&L Runway

Projected cumulative theta income assuming spot pins. Accounts for positions expiring along the way.

■ DTE Traffic Lights

Loading…

■ Collateral Stress Runway

Loading…

■ Assignment / Call-Away Ladder

Loading…

■ Governance & Concentration

Loading…

■ Greek Buckets by Tenor (incl. Vanna / Charm)

Loading…

■ What-If Simulator — Preview Trade Impact

TYPE

DIR

STRIKE

EXPIRY (d)

QTY

IMPACT

—

■ Scenario NAV Cube — Spot × Vol Shocks (7d horizon)

Loading…

■ ATM IV & Realised Vol History

Time series for ATM IV, DVOL, realised vol, IV–RV spread. Requires Deribit historical data feed or Amberdata backfill for institutional-grade history.

Phase 2 — requires historical data warehouse

■ Percentiles & Z-Scores

Percentile ranks over 1m, 3m, 1y, 2y windows for ATM IV, skew, butterfly, term slope, forward vol. Z-score standardisation against rolling means.

Phase 2 — requires historical data warehouse

■ Similar Regime Lookup

Find historical analogues matching current ATM IV percentile, IV–RV spread, skew, and term structure shape. Forward BTC and IV path outcomes after comparable signals.

Phase 2 — requires historical data warehouse

■ Strategy Backtests

Carry, hit rate, drawdown, and tail loss for each strategy template across historical regimes. Strategy-specific signal deployment backtests.

Phase 2 — requires historical data warehouse

■ Metaplanet BIG Strategy — Building Japan’s Bitcoin Standard TSE: 3350 · OTCQX: MTPLF

Bitcoin Income Generation (BIG) Program

“Acquiring bitcoin through the issuance of long-dated yen liabilities… utilising the entire range of capital market instruments… harnessing Japan’s global capital cost advantage.” — Metaplanet Whitepaper, May 2024

Treasury

40,177

BTC · #3 globally

BTC/Share Δ

+5,000%

since BTC standard adoption

BIG Rev FY25

¥8.58B

~$56M · 95% of revenue

Net Cash Rank

#13

of 3,001 JP non-fin cos · $2.34B

Shareholders

216.5K

from 10.9K (Dec 2023) · 20×

2027 Target

210K

BTC · 1% of supply · ~$12B

■ Japan Macro Thesis — The Release Valve

Financial repression is the path of least resistance. Japan has the highest debt-to-GDP globally at 240%. Fiat currency across G7s will be the release valve.

240%

Debt-to-GDP

2.2%

10Y JGB · decade high

3%+

Inflation returning

$7T+

Fixed income at risk

Japan household financial assets: $14.9T (Cash & Deposits $7.6T · Stocks $1.8T · Insurance & Pensions $3.9T). Metaplanet’s listed preferred instruments target this $15T pool.

■ Capital Instruments — Capitalising on Volatility

Instrument	Amount	Market Std	Metaplanet	Edge
Moving Strike Warrants	$1.5B+	8–10% disc	0% disc	+8–10%
Stock + Warrants	$200M+	At market	+5% prem	+5%
MERCURY Preferred	$150M	5–6% yield	4.9% face	Lower cost
MARS Perpetual Pref	Auth	—	<8% · Perp · Senior	Non-dilutive
BTC Income Gen (BIG)	$55M	N/A	Real income	New revenue

Capital raised: 2024 ~$0.13B · 2025 ~$3.25B (common + prefs + warrants) · YTD 2026 ~$0.13B. 555M Plan: ¥770B (~$5.3B) — largest MSW issuance in Japanese capital markets history.

Long-Term HODL

40,177 BTC

Strategic reserve. Never sold. 85% of all BTC held by Japanese public companies. #13 net cash among 3,001 non-financial listed companies in Japan ($2.34B).

Dynamic Collateral

$1B

BTC-backed credit facility. Straight bonds. Monetising the treasury as collateral while retaining full upside exposure. Zero-coupon bond structures.

BTC Income Generation

$55M

FY2025 revenue. Monetising BTC volatility via cash-secured put options. Op. profit +1,695% YoY. Revenue +738%. Growing ~57% per quarter.

Leg 1 — Cash-Secured Put Writing on BTC

Mechanism: Sells OTM cash-secured puts on Bitcoin (via QCP Capital OTC). If BTC stays above strike → pure income. If exercised → buys BTC at strike minus premium.

Roll-up: Dynamically chases rising strikes in bull markets. Oct 2024: $62K → $66K, +5.9 BTC, yield 10.75% → 13.4%.

Edge: Bitcoin VRP (IV 60–100% vs lower realised). ~5–10% of capital in rolling tranches. ¥8.58B FY25 premium revenue.

Leg 2 — Moving Strike Warrants (MSWs)

Mechanism: MSWs reset exercise price daily to prior close — always ATM. EvoFund borrows from MMXX, sells at market, exercises at yesterday’s close, returns new shares.

Advantage: 0% discount (vs 8–10% market standard). Dilution only on upside. Self-reinforcing flywheel.

Scale: 555M Plan — ¥770B (~$5.3B). Prior 210M program raised ~$645M in ~3 months. Daily vol >$920M.

■ The Volatility Flywheel

BTC holdings grow → stock more volatile → put premiums higher → MSW spreads exercisable → more capital → more BTC → repeat. Each revolution increases BTC-per-share.

BTC Yield

568%

Op Profit

+1,695%

30d HV

133.9%

mNAV

3–9.5×

⚠ Key Risk Vectors

Crash assignment: BTC −40%; puts exercised above spot. Short-gamma fat-tail.

IV compression: DAT sellers suppress vol. 13%+ yields → 4–6%.

mNAV collapse: 5–9× → 1× = MSW issuances become dilutive.

MTM: ¥102.2B unrealised loss FY25. Avg cost ~$107K vs ~$68K spot.

Governance: EvoFund ↔ MMXX (Gerovich). Concentration risk.

■ Capital Market Roadmap — “2028: Year 0 for Bitcoin in Japan”

2024

Bitcoin = speculative asset. Capital raised: ~$0.13B via common stock.

2026

Bitcoin = niche corporate asset. Amplified BTC equity + lower-vol preferred instruments. ~$3.5B raised.

2028

Bitcoin = regulated financial instrument. Spot ETF approval + big bank adoption. Futures, options, structured products.

2030+

Bitcoin = capital markets collateral asset, monetary base layer. BTC-backed fixed income products at scale.

Bottom Line

Metaplanet is not merely holding Bitcoin — it is building a full-stack capital markets engine around it. BIG sells tail-risk insurance (put premiums in benign markets, BTC acquisition if exercised). MSWs convert equity volatility into capital at zero discount. MARS/MERCURY preferred instruments tap Japan’s $14.9T household asset pool. The 210,000 BTC target by 2027 would represent 1% of total supply. The strategy compounds BTC-per-share regardless of price direction — provided volatility remains elevated. Fragility: a low-vol, declining-BTC regime where premiums thin, mNAV collapses, and the flywheel reverses.

₿ BTC Asia — Intelligence Hub Advisory · Pre-Deal

Philippines Bitcoin Lending & Treasury Company

Philippines Bitcoin lending and treasury vehicle — combining the Ledn model (BTC-collateralised lending, yield generation) with Metaplanet-style balance sheet accumulation. Capturing structural remittance tailwinds, high crypto adoption, and a clear first-mover gap in ASEAN institutional Bitcoin infrastructure.

LAST UPDATED · Apr 14 2026

Data: public filings + industry reports

Asia Precedent

40,177 BTC

Metaplanet (Japan TSE:3350) — 3rd largest corporate BTC holder globally. First non-US entrant into top-3. $4.2B cost basis. Targets 210k BTC by 2027.

PH Crypto Adoption

#8 GLOBALLY

Chainalysis Global Crypto Adoption Index 2024. ~11M crypto users today · 12.79M forecast by 2026 · 10.86% penetration. Higher than US, UK, Germany.

Remittance TAM

$38.3B/YR

PH inbound remittances 2024 — one of the world's largest per-capita. Structural demand for Bitcoin-rail settlement via Lightning Network. +3% YoY growth.

Crypto Transaction Volume

~$40B/YR

Annual PH crypto transaction value per SEC data. Active, tech-savvy population. Play-to-earn (Axie Infinity) pioneered early adoption.

Strategic Reserve Bill

10,000 BTC

House Bill 421 (Aug 2025) — proposes BSP buys 2,000 BTC/yr × 5 yrs, 20-year hold. Marcos administration publicly pro-Bitcoin.

CASP Framework

₱100M MIN

PH SEC MC 4/5-2025 in force. ~$1.8M min. paid-up capital for licensed crypto operators. High barrier = regulatory moat for early licensees.

Philippines Macro Environment

CPI Inflation

–

BSP Key Rate

–

M2 Growth YoY

–

GDP Growth YoY

–

Loading PHP/USD…

Philippines FX & BTC Charts

USD / PHP

Loading chart…

PHP / BTC

Cost of 1 BTC in Pesos — rising line = peso debasement

Loading PHP/BTC…

🇵🇭 Philippines Crypto & Remittance Ecosystem

BSP-regulated · $38B/yr remittance · 11M+ crypto users

CRYPTO EXCHANGES & WALLETS

Coins.ph	Largest PH platform · exchange + wallet · remittance + bill pay · millions of users
PDAX	BSP-regulated · BTC/ETH in PHP · institutional · powers GCrypto (GCash)
Maya PH	Digital bank + crypto (ex-PayMaya) · Voyager-backed
GCrypto	Crypto inside GCash (76M+ users) · powered by PDAX

BLOCKCHAIN REMITTANCE

BloomX	Crypto-powered cross-border · Lightning Network settlement
Moneybees	OTC cash ↔ BTC via physical locations · working inventory
XenRemit	Crypto + remittance + payments · BTC settlement rail

FINTECH & WEB3 INFRASTRUCTURE

TopJuan	TopWallet · hybrid crypto + financial services
DA5 / SurgePay	Digital wallet · crypto + remittance + bill pay · nationwide agents
Torus Labs	Web3 infra · keyless wallet login · PH-founded

GLOBAL EXCHANGES (PH POPULAR)

Binance	World's largest · P2P PHP on-ramp · most popular with Filipino traders
KuCoin / Bybit	Advanced trading + altcoins · used alongside local for depth

BANKS & INSTITUTIONAL

UnionBank	Blockchain adopter · PDAX partner · custody/tokenisation · no BTC on BS
BSP	Central bank · regulates VASPs · HB 421: 10,000 BTC strategic reserve
SEC PH	CASP framework · ₱100M min capital · regulatory moat
PSE	Zero crypto-native listings · first-mover gap for BTC treasury co

OFW REMITTANCE — GLOBAL

Western Union	100k+ agents globally · cash + bank + GCash
MoneyGram	Fast transfers · cash pickup + bank deposit
Ria Money	Competitive fees · N. America + Middle East

OFW REMITTANCE — DIGITAL

Remitly	Built for OFWs · Express/Economy · GCash/Maya direct
Wise	Real exchange rate · transparent fees · cheapest option
WorldRemit	Bank + cash + mobile wallet · fast delivery
Xoom (PayPal)	Fast digital · popular US→PH corridor
TapTapSend	Low-fee UK/Europe/US transfers

OFW REMITTANCE — FILIPINO-OWNED

iRemit	Largest PH-owned · 1000s of payout locations · door-to-door
Metrobank Remit	Metrobank subsidiary · banking-tied products
BPInoy (BPI)	Direct bank transfer · global partnerships
PNB	International branches · OFW-focused products
Optimum Exchange	Partner houses in Middle East + Asia

CASH PICKUP NETWORKS (PH SIDE)

Palawan Express	Fast low-cost · 1000s of rural branches · Pera Padala
Cebuana Lhuillier	Highly trusted · real-time payouts · financial services
M Lhuillier	24/7 in some areas · loans + pawn · nationwide

BTC HOLDINGS INTELLIGENCE

Coins.ph	Likely significant custodial BTC · ❌ no treasury disclosure · ❌ no on-chain transparency
PDAX	Exchange wallets (custodial) · ❌ no proprietary reserve disclosed
Moneybees	Working BTC inventory for OTC liquidity · ❌ no audited holdings
Nichel Gaba	PH crypto leader · infrastructure builder · influence not holdings
Wei Zhou	Ex-Binance CFO · PH-connected · no confirmed BTC amounts

No Philippine MicroStrategy: BSP regulates tightly · PH corporates favour cash/real estate · crypto adoption is retail + remittance-driven, not institutional · wealthy Filipinos use offshore (SG/HK/Dubai) · PSE has zero crypto-native listings · clear first-mover gap for BTC lending + treasury company.

Signal: on-chain flows from PH exchanges are the real due diligence edge

Comparable Listed BTC Treasuries — Asia Focus

Source: bitcointreasuries.net · Apr 2026

Company	Ticker	Region	Description	BTC Held	Mkt Cap	Price	1M %	3M %	12M %	YTD %
Strategy	MSTR	🇺🇸 USA	Largest corporate BTC holder globally	762,099	—	—	—	—	—	—
Twenty One Capital	XXI	🇺🇸 USA	Tether/SoftBank-backed BTC vehicle	43,514	—	—	—	—	—	—
Asia-Pacific BTC Treasury Companies
Metaplanet ⭐	3350 (TSE)	🇯🇵 Japan	Asia’s #1 · first JP Bitcoin treasury company	40,177	—	—	—	—	—	—
Boyaa Interactive	0434.HK	🇭🇰 Hong Kong	Online gaming → BTC treasury pivot · Web3	3,351	—	—	—	—	—	—
Canaan Inc	CAN	🇸🇬 Singapore	ASIC miner mfg (Avalon) · self-mining · 266 MW	1,808	—	—	—	—	—	—
BitFuFu	FUFU	🇸🇬 Singapore	World’s largest cloud mining platform · 623K users	1,709	—	—	—	—	—	—
YDKG	YDKG	🇨🇳 China	China-listed BTC treasury holder	819	—	—	—	—	—	—
SOS Limited	SOS	🇨🇳 China	AI + blockchain tech · BTC mining + treasury	803	—	—	—	—	—	—
Convano Inc	6574.T	🇯🇵 Japan	Nail salon chain → BTC treasury strategy	763	—	—	—	—	—	—
Bitmax	377030.KQ	🇰🇷 Korea	Korea’s largest listed BTC treasury	251	—	—	—	—	—	—
PH BTC Co (TBD) 🎯	TBD	🇵🇭 Philippines	First ASEAN BTC treasury vehicle · advisory	TBD	—	—	—	—	—	—

Asia-Pacific now has 8+ listed BTC treasury companies across Japan, Hong Kong, Singapore, China, and Korea. ASEAN (Philippines, Indonesia, Thailand, Vietnam, Malaysia) currently has zero — clear first-mover gap. Metaplanet is the only non-US entrant in the global top 3.

Thesis

Asia Pacific has one publicly-listed Bitcoin treasury company (Metaplanet). Southeast Asia has zero. The Philippines sits at the intersection of three structural tailwinds:

Adoption: #8 globally, ~10% population penetration, 12.8M users by 2026
Tailwind: $38B/yr remittance TAM, prime use-case for Lightning-rail settlement
Regulation: CASP framework in force · Marcos administration explicitly pro-Bitcoin · HB 421 proposes 10k BTC strategic reserve

A Manila-listed treasury vehicle would be first-mover in ASEAN and capture capital flows from the crypto-native investor base that today has no locally-listed BTC equity exposure outside of MSTR proxy.

Deal Structure

Advisory phase — structure TBD. Likely vehicle candidates:

Direct PSE IPO (Philippine Stock Exchange)
Reverse merger into existing listed shell
Backdoor listing via SPAC precedent (Twenty One model)
Offshore dual listing (Singapore SGX or HKEX + PSE)

[ Deal structure details to be filled as advisory progresses ]

Regulatory Environment

Current status (April 2026):

SEC MC 4 / MC 5-2025 — CASP rules in force · ₱100M min. paid-up capital · physical presence required
BSP VASP license — moratorium lifted Sept 2025
HB 421 — proposes Strategic Bitcoin Reserve (10,000 BTC target, 20-year hold)
Cagayan CEZA Crypto Valley — designated crypto economic zone since 2018
Tax treatment — capital gains reporting required, no specific crypto CGT rate

Investors — Target Profile

Target investor profile to approach for this deal:

Bitcoin-native funds (e.g. Paradigm, Pantera, 10T, Multicoin)
Asia-focused crypto funds (Dragonfly, HashKey, Animoca)
Japanese institutional allocators (post-Metaplanet precedent)
Philippine family offices with crypto exposure
Strategy-correlated US funds looking for Asia diversification

[ Investor list + cross-linking to Investor Intelligence page pending ]

Research & Documents

Working documents, due diligence materials, and reference research will be linked here as the deal progresses.

[ Document vault TBD ]

Sources: Metaplanet Q1 2026 disclosure (April 2026); Bitcoin Treasuries leaderboard; Chainalysis Global Crypto Adoption Index 2024; PH SEC Memorandum Circulars No. 4 & 5, Series of 2025; BSP VASP regulatory framework; House Bill 421 (Aug 2025); PH SEC CASP statistics. Macro figures are point-in-time and should be refreshed quarterly as conditions evolve.

Asia BTC + Crypto News

— JP · KR · HK · SG · CN · TW · PH · TH · IN · VN + RSS

Loading Asia news…

REVENUE ($M)	2026	2027	2028	2029	2030
P&L SUMMARY ($M)

Bitcoin · BTC/USD · CoinGecko

–

Loading…

24H High

–

24H Low

–

24H Volume (USD)

–

Market Cap

–

Circulating Supply

–

All-Time High

–

24H Range –

52-Week Range –

Technicals · Daily yfinance · 13mo history

Vol 30D

–

Vol 90D

–

RSI 14

–

EMA 21

–

EMA 50

–

EMA 200

–

Bitcoin Headlines —

16 tickers + 12 RSS feeds · 7-day window

Loading Bitcoin headlines…

Memory Bank

Loading…

By Category

To-Do

–

Agenda

Loading…

H100 Advisory

Hours by Category

–

Expenses by Category

–

Hours Logged

Expenses

Contacts

Network · Relationships

API Usage

OpenRouter · loading…

Today

–

MTD

–

All Time

–

Calls MTD

–

By Model — MTD

Loading…

Daily Spend — Last 14 Days

Loading…

Investment Intelligence

Pipeline · Life Sciences · Oxford · Neuro · HMA

Loading…

🧬 Life Sciences & Healthcare Headlines

Loading headlines…

🎓 Oxford Ecosystem

Loading…

🧬 Etcembly / AI Drug Design

Loading…

🌍 VX Pharma / Africa

Loading…

Investment Intelligence · Neurotechnology

The Brain Is the Next Great Platform

Aging demographics, maturing closed-loop implants, AI-driven neuroimaging, and BCIs crossing from research into revenue — neurotech is compounding at double-digits into a decade where the clinical and reimbursement tailwinds are structural, not cyclical.

TAM Today

$15–19B

2025 · broad neurotech

TAM 2035

$47–67B

devices alone: $67B

Blended CAGR

11–13%

devices 13.5% · BCI 16–18%

Geography

NA >40%

APAC fastest-growing

$33B broad market by 2031

$100Bs US annual neuro disease cost

2x projected device revenue by 2035

10yr compounding window

■ Macro Thesis

Four tailwinds converging in the same decade

Demographics. Western and East-Asian populations are aging into the peak decade for Parkinson's, Alzheimer's, stroke and chronic pain — the exact indications the implants treat.

Adaptive closed-loop systems work. Next-gen DBS senses neural state and titrates stimulation in real time. Clinical results in Parkinson's are strong enough that device players now lead one of the fastest-growing indication segments in medtech.

BCIs crossing the revenue line. Invasive, minimally invasive venous, and non-invasive EEG platforms are each moving past first-in-human into commercial trials. The category is small today but growing at 16–18% CAGR.

Neuroimaging decentralising. Portable MRI and AI-triaged CT are moving stroke and aneurysm diagnosis out of tertiary centres into EDs and ICUs — a mid-teens CAGR category on its own.

■ Why Payers Reimburse

Implants that keep patients out of institutions pay for themselves

US neurological disease costs run into the hundreds of billions annually when direct medical costs and lost productivity are combined. The cost drivers that move payer math:

→ Readmissions for uncontrolled pain & spasticity
→ Institutional care for dementia & Parkinson's
→ Stroke-related disability & long-term rehab
→ Falls, fractures, aspiration events in the frail

Investor takeaway: high-ticket implants are reimbursable when they keep patients ambulatory, continent and cognitively functional for longer — that's the entire neuromod investment case in one sentence.

■ Subsectors to Track

Eight theses a neurosurgeon-investor should build

ranked by current revenue weight

01 Neuromodulation

WORKHORSE

SCS, DBS, vagus, sacral and peripheral nerve stimulation. Bulk of current device revenue. Parkinson's via adaptive closed-loop DBS is the fastest-growing indication.

DBS · SCS · PNS · VNS · SNS · TMS · FUS

02 Neurotech Devices (aggregate)

$19B → $67B

All device categories combined. ~$19B in 2025, $67B by 2035 at 13.5% CAGR. Strong hospital demand plus expanding ambulatory use for chronic pain and movement disorders.

MONITORING · PROSTHETICS · SENSING WEARABLES

03 Brain-Computer Interfaces

16–18% CAGR

Smallest revenue today, among the fastest-growing. Invasive implants, minimally invasive venous systems, and non-invasive EEG for communication and rehab. Highest optionality slice of the stack.

INVASIVE · VENOUS · NON-INVASIVE EEG

04 Neurodiagnostics & Imaging

MID-TEENS CAGR

EEG/MEG analytics, portable MRI, AI imaging. Moving stroke and cognitive-decline triage out of tertiary centres into EDs, ICUs and point-of-care. Portable MRI alone is a mid-teens CAGR category.

PORTABLE MRI · EEG/MEG · AI TRIAGE

05 Neuroprosthetics

STRATEGIC

High-cost, low-volume today. Limb loss, spinal cord injury, sensory and vision restoration. Shares investors and core tech with BCI — watch for inflection as reimbursement pathways mature.

MOTOR · SENSORY · VISION · SPINAL · CRANIAL

06 Neuromonitoring & Wearables

CONSUMER EDGE

EEG headsets, ICU and peri-operative monitoring, consumer-grade sleep and focus wearables. The bridge from hospital-grade neurotech to everyday clinical and wellness use.

ICU · OR · AMBULATORY · CONSUMER EEG

07 Neurorehabilitation

SOFTWARE-LED

VR, exoskeletons, gamified rehab software. Post-stroke and spinal cord injury recovery. Software economics on top of traditional PT — capital-light, data-rich, outcome-measurable.

VR · EXOSKELETON · GAMIFIED PT

08 Cognitive & Neuropsychiatric Devices

EMERGING

Depression, ADHD, addiction — device-led therapy where pharma has either stalled or created side-effect burden. TMS and peripheral stim already reimbursed; prescription digital therapeutics behind.

DEPRESSION · ADHD · ADDICTION · PTSD

+ Neuromorphic & Neuro-Inspired Hardware ENABLING LAYER

Brain-inspired compute architectures for on-device AI inference. Not a clinical segment yet — watch as an upstream enabler: the devices above all eventually need low-power, event-driven silicon to run closed-loop models at the patient.

■ ASC Market — The Macro Opportunity

HOPD Cost/Case

$7,716

hospital outpatient avg

ASC Cost/Case

$3,157

>50% lower · same outcomes

Medicare Savings

$4.2B

annual · CA alone >$630M

Patient Satisfaction

>90%

8M+ ASC patients/yr

Site-of-Service Cost Arbitrage

Consumer saving per procedure (UHG)~$684

Cataract coinsurance: HOPD → ASC$496 → $195

Colonoscopy coinsurance: HOPD → ASC$186 → $89

Lower inpatient admission and death rates vs HOPDs after risk adjustment. Subject to state licensure, Medicare certification, voluntary accreditation (JCAHO, AAAHC, AAAASF, AOA).

ASC Market Infrastructure

Total US ASCs11,352

Market size (2022 → 2027)$43.1B → $56.3B (5.5% CAGR)

Physician-only owned60%

Avg ORs per facility~3.1

National chains own a small fraction. Fragmentation + industrial logic for scale = classic roll-up opportunity. Neurosurgery ASC penetration near zero.

Commercial Growth Dynamics

Volume migration. $500B inpatient + $90B outpatient surgical TAM. ~$60B (3.6M cases) clinically suitable for outpatient migration. Neuro penetration near zero.

Case mix upgrade. Ortho/spine/neuro/pain expected to contribute ~160% of migrating volume. Higher reimbursement, high-margin procedures.

Demographics. US population ≥65 growing 15% → 24% by 2060. Rising degenerative spine, cerebrovascular, and neurosurgical indications.

Regulation. CMS expanding ASC Covered Procedures List yearly. Key advisory bodies recommend paying ASCs for any procedures not requiring overnight stay.

Benchmark — Surgery Partners (Pure-Play ASC)

Facilities139 ASCs + 18 hospitals, 32 states

Scale~5,000 physicians · ~650K patients/yr

TTM Revenue (Q3 2023)~$2.7B

Adj. EBITDA / margin~$417M / 15%

EBITDA CAGR (2017→2023)Mid-teens %

Acquisitions>$1B at ~8× pre-synergy

MSK % of net revenue~50%

Quality10/10 benchmarks · 4–5★ CMS

🧠 Neuroscience / Neurotech Headlines

14-day window · sub-sector classified · synced with Overview feed

Loading…

Segment	2023	2024	2025	2026	2027
Spine	12.9	15.9	19.4	20.0	20.6
Endovascular	—	0.6	1.2	2.9	3.0
Radiosurgery	—	—	1.6	4.1	4.2
Injections/RF/Stim	2.0	3.2	3.3	3.4	3.5
Total ($M)	14.9	19.7	25.5	30.3	31.2
Net Income	3.3	3.7	7.6	11.7	11.8
Net Margin	22%	19%	30%	39%	38%

Member	Stake	Role / Notes
Dr. Douglas B. Moreland	20%	General Neurosurgery · MI Spine
Dr. Michael Stoffman	15%	General Neurosurgery · MI Spine
Dr. John G. Fahrbach IV ⚠	15%	MI Spine · Cranial Surgery — stake to be confirmed
Dr. Elad I. Levy	10%	Chairman · Cerebrovascular · MI Spine · Cranial
Cournyea	10%	Non-physician · Does not perform cases
Member F	10%	Physician · Name TBC
Member G	10%	Physician · Name TBC
Member H	5%	Physician · Name TBC
Member I	5%	Physician · Name TBC

NI Multiple	2024 NI	2025P NI	2027P NI
8×	$30	$61	$94
10×	$37	$76	$118
12×	$44	$91	$142
15×	$56	$114	$177

	2023	2024	2025	YoY (24→25)	CAGR (23→25)
Revenue	$15,532,000	$20,240,000	$26,000,000	+28.5%	+29.4%
EBITDA	$3,627,000	$5,379,000	$6,500,000	+20.8%	+33.8%
EBITDA Margin	23.3%	26.6%	25.0%	−1.6pp
Implied OpEx	$11,905,000	$14,861,000	$19,500,000	+31.2%

Service Line	2023	% Mix	2024	% Mix	2025	% Mix	YoY
Spine	917	32%	1,132	31%	1,169	26%	+3%
Endovascular	7	0%	205	6%	280	6%	+37%
Pain / Injections	1,970	68%	2,313	62%	2,625	59%	+13%
Interventional Radiology	3	0%	52	1%	380	9%	+631%
Total	2,897	100%	3,702	100%	4,454	100%	+20%

	2023	2024	2025	FORECAST ONLY
Revenue ($M)	14.9	15.5	+4%	19.7	20.2	+3%	25.5	26.0	+2%	30.3	31.2
EBITDA ($M)	—	3.6	—	—	5.4	—	—	6.5	—	—	—
EBITDA Margin	—	23%		—	27%		—	25%		—	—
Net Income ($M)	3.3	—		3.7	—		7.6	—		11.7	11.8
Net Margin	22%	—		19%	—		30%	—		39%	38%

Comp	Type	EV	Revenue	EBITDA	Margin	EV/EBITDA	EV/Rev
Surgery Partners (SGRY)	Public · 200+ ASCs	$7.7B	$3.3B	$526M	15.9%	14.6×	2.3×
USPI (Tenet subsidiary)	Subsidiary · 500+ ASCs	—	$1.28B	$492M	38.6%	12.5× acq	—
AmSurg / Ascension	M&A · 250+ ASCs	$3.9B	—	—	—	est ~10–12×	—
SCA Health / Atlas offer	Bid · single center	$90M	$26M	$6.5M	25.0%	13.8×	3.5×
Private ASC transaction ranges (Scope Research, 2024–25)	—	5–10×	—

	Bear	Base	Bull
Metric Year	2025A	2026P	2027P
Revenue ($M)	26.0	30.3	31.2
EBITDA ($M)	6.5	9.0	11.8
EV/EBITDA	7.0×	10.0×	13.0×
Enterprise Value	$45.5M	$90.0M	$153.4M
Less Debt	(6.0)	(6.0)	(6.0)
Equity Value	$39.5M	$84.0M	$147.4M
30% Stake Value	$11.9M	$25.2M	$44.2M

Channel	Jan	Feb	Mar	Apr	May	Jun	Jul	Aug	Sep	Oct	Nov	Dec	Total
HCA	85K	142K	152K	196K	139K	147K	116K	17K	0	0	0	0	$993K
AWC	331K	62K	91K	1K	14K	0	0	0	0	0	0	0	$500K
Direct	4K	98K	76K	51K	19K	22K	76K	64K	68K	209K	59K	43K	$789K
Total	420K	303K	325K	250K	182K	169K	192K	81K	69K	210K	59K	43K	$2.30M

Class	Shares (FD)	%	Cash In
Common Stock	13,948,032	85.06%	$9.35M
CN Notes (convertible)	—	—	$7.88M
Options/RSUs	1,765,773	10.77%	—
EIP Available	684,227	4.17%	—
Total (FD)	16,398,032	100%	$17.22M

Product	Year 1	Year 2	Year 3	Year 4	Year 5
NeoMatriX Wound Cover	$1.6M	$3.2M	$21.2M	$29.1M	$35.4M
NeoMatriX Nerve Cover	—	$3.3M	$17.6M	$27.0M	$55.0M
Tendon Cover	—	—	$2.9M	$6.0M	$12.2M
Bone + Particulate	—	$30K	$1.1M	$10.6M	$20.3M
Total	$1.6M	$6.5M	$42.9M	$72.6M	$122.9M

Flag	Detail
Revenue collapse	Revenue fell 90% from Jan ($420K) to Dec ($43K). Both HCA and AWC went to $0. Meeting notes say “$5M confident this year” but H2 run rate ex-Oct is $600–840K annualised.
Valuation inconsistency	Meeting notes: $49–55M post for $5M. Shareholder report: $60M at $3.69/share for $3.2M. Investor deck: $25M Series B (valuation unstated). Meeting notes also mention $100M for B2. Four different figures across three documents.
Cash position mismatch	Meeting notes claim “$1M cash + $800K AR.” Shareholder report (31 Dec 2025) shows $462K cash + $389K AR = $851K total. The $1M cash claim requires ~$540K raised in Q1 2026.
HCA not explained	Meeting notes focus on AWC going bankrupt but don’t mention HCA. HCA was $993K in 2025 (43% of revenue) and went to $0 after August. This is a larger revenue loss than AWC ($500K).
Negative equity	Total equity is ($73,848). Retained earnings: ($20.2M). AP ($1.05M) exceeds cash by 2.3×. Company has burned through $20M+ of investor capital with only $2.3M peak annual revenue.
AWC overstated	Meeting notes say AWC was “$2.5M revenue.” Actual 2025 AWC revenue was only $500K (and declining — $331K Jan, $14K May, then $0). The $2.5M may be historical or aspirational.
Projection extremity	5-year projection: $1.6M → $122.9M (77×). Requires 4 simultaneous FDA clearances, commercial launches, and aggressive market penetration. Historical execution: $2.3M revenue after 11 years since founding.
Round fatigue	B Series round has been open for extended period during biotech down-round environment. Shareholder letter now asks for just $3.2M (down from $25M). JT acknowledged communication issues. Convertible notes total $7.88M — conversion overhang unclear.
Medicare headwind	Dec 2025 Medicare reimbursement change acknowledged but unquantified. Could materially impact the wound care revenue base — needs diligence.
Capital raised discrepancy	Website says “over $13 million in startup capital.” Cap table shows $17.22M raised ($9.35M common + $7.88M convertibles). Likely website is outdated.

Company Watchlist

Loading…

Investment Intelligence · Confidential

Deal Flow Pipeline

COMPANY ↕	COUNTRY ↕	STAGE ↕	SECTOR ↕	PRIORITY ↕	STATUS ↕	RAISE ↕	PRE-$ ↕	MOIC ↕	ROI ↕	SCORE ↕	·
Loading pipeline…

💼 Investor Intelligence

Loading investor data…

All Investors

Type:

Tags:

Loading…

📊 Comparable Transactions

Loading comps…

🍄 Mycofeast

No news cached — click ↻ Refresh News

Developer

Build reference · v257 · v257

System Architecture

📱 Telegram
@agent00X_bot

⟷

⬡ AGENT00
Orchestrator

⟷

🌐 Web Dashboard
Flask :8080

Sub-Agents

🕐 AGENT01
H100 Advisory

📊 AGENT02
Trading Intelligence

₿ AGENT03
Bitcoin Intelligence

🔬 AGENT04
Inv. Intelligence

📡 AGENT05
Market Data

💰 AGENT06
Personal Finance

🧠 Memory
Soul + KB

🗄 SQLite
agent00.db

Files: AGENT00.py · AGENT01_time/flow/state.py · AGENT02_trading.py · AGENT03_btc/pdf.py · AGENT04_lifesci.py · AGENT00_web/db/memory/system.py

External APIs

📈 yfinance
OHLCV · 15m delay

🤖 OpenRouter
LLM · gpt-4.1-mini

🌤 Open-Meteo
Weather · London

🔍 Brave Search
News · LS digest

📰 NewsAPI
Headlines

Stack

Runtime: Python 3.9 · Mac Mini (honeybees-Mini) · always-on
Web: Flask · port 8080 · single-file · vanilla JS · dark theme
Database: SQLite — agent00.db
DB tables: setup_alerts · trade_results · eod_summaries · memory · todos · agenda · h100_hours · h100_expenses · event_log · ls_watchlist · ls_deals · ls_news_cache · ls_digests · contacts · contact_tags · pfin_assets · pfin_expenses · pfin_expense_tags · pfin_wealth_history · pfin_company_accounts · investors · deal_investors · investor_tags · trade_overrides · strategy_changelog · decision_journal · comps · deal_reviews · options_positions · treasury_config
LLM: OpenRouter — openai/gpt-4.1-mini default · anthropic/claude-sonnet-4-5 for deck import · configurable via OPENROUTER_MODEL in .env
Process mgmt: launchd — com.agent00.plist
Data: yfinance (15-min delayed) · Open-Meteo weather

Files

AGENT00.py — orchestrator, Telegram, command router
AGENT01_time.py — AGENT01 Time Management · hours & expenses
AGENT01_flow.py — AGENT01 guided Telegram entry flows
AGENT01_state.py — AGENT01 conversation state persistence
AGENT02_trading.py — AGENT02 Trading Intelligence
AGENT03_btc.py — AGENT03 Bitcoin Intelligence
AGENT03_pdf.py — AGENT03 institutional PDF reports
AGENT04_lifesci.py — AGENT04 Investment Intelligence (Life Sciences · Deal Pipeline · Oxford)
AGENT00_memory.py — soul prompt + memory bank
AGENT00_db.py — SQLite read/write layer
AGENT00_system.py — health, backups, cost tracking
AGENT00_web.py — this dashboard
restart.sh — clean restart script
.env — API keys (never committed)

AGENT00 v3 — System Architecture

v3 restructures the platform from separate tools into one integrated intelligence OS. Seven layers, three cross-domain connectors, two-phase data strategy.

Seven System Layers

L7 Decision Layer — unified opportunity engine · ranked actions · capital allocation
L6 Reporting — daily/weekly reviews · strategy health · investment memos · Telegram digests
L5 Analytics — expectancy engine · regime detection · confluence scoring · deal scoring
L4 Alerting — signal generation · deal signals · entity alerts · milestone triggers
L3 UI — cockpit dashboard · domain views · charts · action cards
L2 Memory — decision journal · interaction log · rationale capture · audit trail
L1 Data — SQLite · trade schema · deal schema · investor schema · controlled vocabs

Three Cross-Domain Connectors

Entity Graph — company/investor/contact canonical IDs · headline entity matching · auto-surfaces relevant intel on deal pages
Unified Opportunity Ledger — every actionable item (trade, deal, meeting, milestone) feeds one ranked queue
Decision Journal — every decision logged with rationale, confidence, outcome · learning loop

28-Day Data Protocol

Schema Freeze Rules

FROZEN: strategy names (ICT/AMD50/LIQVOL/MEANREV/ORB_5M/ORB_15M) · asset codes (GC/ES/BTC/CL) · direction (LONG/SHORT) · result_status · tier values · kill zone names
ALLOWED: additive new columns with defaults · new controlled vocab values · new tags
FORBIDDEN: renaming enum values · changing field meaning · deleting tables · changing signal logic without version bump

Minimum Viable Dataset (28-Day Target)

MES	30 target · 15 min
MGC	25 target · 12 min
BTC	20 target · 10 min
MCL	15 target · 8 min
Total	90 target · 45 min

Phase 1 vs Phase 2 Gating

Phase 1 — Pre-28-Day (Active Now)

✅ Clean data capture — all 17 setup_alerts fields populated on every signal
✅ Regime auto-detection — ATR → volatility, ADX → trend, calendar → macro
✅ Drawdown controls — $700 daily cap enforced at signal creation + suppress-check
✅ R-multiple auto-computed on every close via _enrich_trade_result()
✅ R-multiple backfill — all existing closed trades retroactively computed (pnl/350) at startup
✅ Day of week backfilled on all existing trades
✅ Controlled vocabularies enforced
✅ Decision journal — 4 API endpoints + Developer pill UI (domain filter, log/edit/delete, outcome tracking, retrospective)
✅ Probability-weighted EV — bull/base/bear probability inputs on deals, weighted MOIC/IRR/equity on one-pager + modal
✅ Deal scoring model — 0-100 composite from 8 weighted components, auto-computed on save, displayed on pipeline + one-pager
✅ Confidence/confluence UI capture — 12-option checkboxes on signal cards + PATCH endpoint (v235)
✅ Decision journal auto-logging — 7 hooks: signal create/close/suppress, auto-close TP/SL, EOD, deal status change (v235)
✅ AGENT02 _log_alert() — 1.5× RR gate + risk caps + v3 field capture at source (v234)

Phase 2 — Post-28-Day (Activates at n≥20 per bucket)

✅ Expectancy engine — per strategy × asset × session (v236)
✅ Edge gating — suppress negative-expectancy strategies (v236)
✅ Dynamic sizing — 0.5×/1×/1.5×/2× based on expectancy tier (v236)
✅ Strategy health dashboard — win rate, expectancy, profit factor, trend (v236)
✅ Kill zone heatmap — session × asset performance grid (v236)
✅ Equity curve + R-multiple distribution charts (v236)
✅ Full Risk Analytics panel (v246.23) — distribution moments, VaR/ES (parametric + historical), payoff asymmetry, ACF + Ljung-Box, regime breakdown, path dependency, rolling Sharpe 20/60/120, equity curve with $/% toggle from $100k start
✅ Regime-specific expectancy breakdown (v237)
✅ Regime-aware edge gating — blocks signals in negative VOL×TREND combos (v237)
✅ MFE/MAE live tracking + analytics — entry/exit quality, capture ratio (v237)
✅ Edge decay detection — rolling 20-trade expectancy (v236)
✅ Confluence score vs win rate analysis (v236)
✅ Confluence backfill — NLP extraction from rationale + editor on closed trades (v237)
✅ Post-trade diagnosis — 4-grade panel (thesis/entry/management/rules 1-5) + loss attribution (v237)
✅ Strategy health report — auto promote/demote/pause/re-test per strategy (v237)
✅ Regime backfill — historical yfinance data to tag existing trades (v237)
☐ Dynamic sizing enforcement — wire sizing_mult into actual position size
☐ Unified opportunity ledger + ranking

Controlled Vocabularies

strategy_name	ICT · AMD50 · LIQVOL · MEANREV · ORB_5M · ORB_15M
asset	GC · ES · BTC · CL
direction	LONG · SHORT
tier	HTF · LTF · SCALP
session	ASIA · LONDON · NY_AM · NY_PM · OVERNIGHT
kill_zone	ASIA_KZ · LONDON_KZ · NY_KZ · POWER_HOUR · NONE
volatility_regime	LOW · NORMAL · HIGH · EXTREME
trend_regime	RANGE · WEAK_TREND · STRONG_TREND
macro_proximity	NONE · SAME_DAY · NEXT_DAY · THIS_WEEK
result_status	WIN · LOSS · BE · SCRATCH · PARTIAL · OPEN
close_reason	TP_HIT · SL_HIT · MANUAL · EOD · TRAILING · TIME_EXIT · DRAWDOWN_HALT
trade_mode	PAPER · LIVE · SIMULATED
bias_state	WITH_TREND · COUNTER_TREND · NO_BIAS · UNKNOWN
structure_state	BOS_UP · BOS_DOWN · RANGE · CHOCH_UP · CHOCH_DOWN
liquidity_condition	SWEPT · BUILDING · THIN · UNKNOWN

Trade Gating Logic (8 Gates)

Gate	Rule	Phase 1	Phase 2
G1 Structural	Meets strategy setup criteria	Active	Active
G2 Risk	R:R ≥ 1.5, within $ caps	Active	Active
G3 Cadence	No duplicate within window	Active	Active
G4 Kill Zone	Inside high-liquidity session	Active	Active
G5 Regime	Correct vol/trend regime	Logged	Active
G6 Confluence	Min confluence score	Logged	Active
G7 Drawdown	Within daily/weekly loss cap	$700/day	Dynamic
G8 Edge	Positive expectancy proven	Off	Active

v3 DB Schema Summary

New Columns on setup_alerts (17)

confidence_score · confluence_score · confluence_components · volatility_regime · trend_regime · macro_event_proximity · macro_event_name · day_of_week · trade_mode · bias_state · structure_state · liquidity_condition · close_reason · timeframe_htf · timeframe_entry · decision_rationale · edge_bucket

New Columns on trade_results (9)

pnl_r_multiple · mfe · mae · close_reason · execution_grade · thesis_grade · rule_following_grade · trade_mode · slippage_estimate

New Tables (5)

trade_overrides · strategy_changelog · decision_journal · comps · deal_reviews · options_positions · treasury_config · pfin_expense_tags

New Columns on ls_deals (v234)

bull_probability · base_probability · bear_probability · composite_score

New Column on comps (v235)

deal_id (INTEGER — optional link to ls_deals)

Post-28-Day Analytics (Phase 2)

Expectancy: (win_rate × avg_win) − (loss_rate × avg_loss) per strategy × asset × session
Dynamic sizing: E[R] < 0 → suppressed · 0–0.15 → 0.5× · 0.15–0.30 → 1× · 0.30–0.50 → 1.5× · >0.50 → 2×
Min sample: 20 per bucket = PROVISIONAL · 50 = CONFIRMED
Edge decomposition: model edge · execution edge · management edge · variance
Edge decay: rolling 20-trade expectancy · flag if <0 for 20 consecutive
Interaction effects: strategy × asset × session × regime cross-tabs
MAE/MFE: entry quality (MAE > 0.5R = late) · exit quality (MFE >> actual R = premature)
~~Deal scoring / probability-weighted EV~~ ✅ shipped v234 (Phase 1)
Unified opportunity ranking: urgency × return × confidence × evidence ÷ effort

Loading agent status…

AGENT02 — Trading Intelligence

Assets: BTC · MGC (Gold) · MNQ · MES · MCL — all micro contracts
Timeframes: 4H · 1H · 15M per asset via yfinance
EMA: 9 / 21 / 50 / 200 — stack alignment, 9/21 crossover detection
VWAP: session VWAP, above/below bias, reclaim/rejection signals
RSI(14): state labelled — OVERBOUGHT / BULLISH / NEUTRAL / BEARISH / OVERSOLD
MACD: line, signal, histogram, trend state (BULLISH_MOMENTUM / FADING / BEARISH_MOMENTUM)
ATR(14): stop = 1.5×ATR · target = 3×ATR · auto-scales per volatility · BTC override: stop = 2.5×ATR · target = 5×ATR (targets ~$350 risk on $10k position)
Bollinger Bands: squeeze detection · expansion with EMA alignment = cleaner trend
Direction filter: LONG_ONLY / SHORT_ONLY / NO_TRADE — hard gate on signal generation
Signal quality score: 0–10 composite based on indicator alignment

Contract Specifications

Display	Internal	Full Name	$/pt	Feed	Trading Hours (UK)
MNQ	MNQ	Micro E-mini Nasdaq-100	$2.00	NQ=F	Sun 23:00 – Fri 22:00 · daily break 22:00–23:00
MES	ES	Micro E-mini S&P 500	$5.00	ES=F	Sun 23:00 – Fri 22:00 · daily break 22:00–23:00
MGC	GC	Micro Gold Futures	$10.00	GC=F	Sun 23:00 – Fri 22:00 · daily break 22:00–23:00
MCL	CL	Micro WTI Crude Oil	$100.00	CL=F	Sun 23:00 – Fri 22:00 · daily break 22:00–23:00
BTC	BTC	Bitcoin (spot, $10k notional)	% return	BTC-USD	24/7 · 365 days — no breaks

P&L formula: pts × multiplier for futures · $350 ÷ stop% dynamic notional for BTC. Price feeds use CME full-size symbols (NQ=F / ES=F / GC=F / CL=F) — P&L scaled to micro sizing. Note: MGC = $10/pt (10 oz × $1/oz) · MCL = $10/pt (100 bbl ÷ 10)

⚡ Risk Management Rules

Asset	Max Risk	Max Stop	Approach
MGC	$350	35pts	Hard cap enforced at signal creation · $10/pt · stop > 35pts rejected
BTC	$350	3.5%	AGENT02_trading.py line 407 · stop = 2.5×ATR (vs 1.5× other assets) · target = 5×ATR · $10k notional · ~3.5% stop = ~$350 risk
MNQ	No cap	ATR-based	$2/pt · ATR stop auto-scales · typical risk $150–400
MES	No cap	ATR-based	$5/pt · ATR stop auto-scales · typical risk $200–500

Signal cap: 1 active signal per asset per strategy (ICT/AMD50/LIQVOL/MEANREV/ORB_5M/ORB_15M each get independent slot) · race condition protected with BEGIN IMMEDIATE DB lock before cap check · Risk caps: MGC max 35pts ($350) · BTC 3.5% of notional (~$350) · MNQ floor $350 min · all caps enforced in _log_alert() in AGENT02 + both Flask endpoints · BTC P&L: Notional = $350 ÷ stop% · dynamic per trade · L always ~$350 · W scales with R:R · cap check uses stop% > 3.85% (was ×2000 — fixed v229) · EOD time guard: run_eod_summary() blocked before 20:00 UK via pytz — prevents false morning runs on server restart

ICT Methodology

FVG — Fair Value Gaps: imbalance zones left by impulsive moves
Order Blocks — last candle before impulsive move, acts as S/R
Liquidity Sweeps — stop hunts above highs / below lows before reversal
BOS/CHoCH — Break of Structure / Change of Character
Premium/Discount — above/below 50% of HTF range
Three tiers: HTF (4H/Daily, R:R ≥2.0) · LTF (1H/15M, R:R ≥1.5) · SCALP (15M, R:R ≥1.0)
Highest probability: FVG + VWAP + EMA pullback + MACD confirm + liquidity sweep

LIQ+VOL — Intraday Liquidity & Volatility (Jadecap)

Markets raid orders at Asian/London range extremes and prev-day H/L then reverse. Wait for sweep → zero follow-through → closure back through → enter. SMT (NQ≠ES) confirms. Win rate ~35% — edge is in management. signal → TRADING:LIQVOL:LATEST

Mean Reversion — Right Side of the V (Lance Breitstein)

Stable boring assets making fast/large moves on no news = out of equilibrium. Score 7 variables (size, ROC, streak, volume, boringness, market cap, MA distance). Trade A+ only (80+/100). NEVER front side. Target = 20MA (equilibrium). signal → TRADING:MEANREV:LATEST

AMD 50% Range Rebalance — Power of Three

Strategy from the strategy author (Chart Fanatics) — $2.3M prop firm payouts. Core thesis: price always redelivers to 50% EQ of the dealing range (Power of Three / AMD model). Works on MNQ (primary) and BTC (daily timeframe). Auto-creates signal when score ≥7/9 + R:R ≥1.5. signal Scan → TRADING:AMD50:LATEST · TRADING:TRADE:{id} on close.

Rules (SHORT)

Price in premium — above 50% EQ of dealing range
Daily PO3 — wick above previous day high (stops run)
H4 PO3 — H4 sweep at 10am EST window
SMT divergence — NQ new high, ES fails to confirm
Inversion FVG — previous support breaks, retap as resistance
Target — 50% EQ (base hit). Trail if volume confirms.
BE — aggressively when H1 candle flips bearish

Key Concepts

AMD — Accumulation → Manipulation → Distribution (Power of Three)
Fractal — same PO3 logic on Daily → H4 → H1 → align all three
Session — 9:15–11:30 EST primary · 10am manipulation reversal key time
SMT — Smart Money Technique: NQ/ES divergence is entry confirmation
50% only — never fight to catch extremes · base hit mentality
4-week gate — track in AGENT00, patterns emerge after 28+ trades

Kill Zones & Session Rules (v240+)

Source of truth: ICT_KILLZONES_ET in AGENT00_web.py and duplicated in AGENT02_trading.py. If these change, both files must be updated.

CME Futures (MNQ · MES · MGC · MCL) — Mon–Fri only

Zone	ET	UK (BST)	Rules
London	02:00–05:00	07:00–10:00	ICT entries allowed · Rule 3 force-close = max(05:00 ET, fill+60m)
NY Open	08:30–11:00	13:30–16:00	ICT entries allowed · Rule 3 force-close = max(11:00 ET, fill+60m)
NY PM	14:00–15:00	19:00–20:00	Observe-only · Rule 2: rejects unless `pm_override=true` from web UI · AGENT02 always rejects (no override channel)
Asia	20:00–24:00	01:00–05:00	ICT entries allowed · Rule 3 force-close = max(24:00 ET, fill+60m)
Last Hour	15:30–16:30	20:30–21:30	Rule 4: all non-BTC `api_signal_create` rejected with `LAST_HOUR_BLOCKED`
NY Close	16:30	21:30	⛔ EOD hard close — all non-BTC OPEN/ALERTED positions force-closed 21:30–22:30 UK
Weekend	Sat–Sun	—	⛔ All CME futures signals blocked · BTC only

Rules — Enforced in Both api_signal_create and AGENT02 _log_alert

Rule 1 — Kill Zone Entry Gate: ICT signals only permitted inside London/NY Open/NY PM/Asia windows · rejects with KILLZONE_GATE_BLOCKED · BTC exempt (v240.2)
Rule 2 — PM Observe-Only: NY PM (14:00–15:00 ET) rejects ICT unless pm_override=true · web UI shows purple override button with confirm dialog · AGENT02 has no override channel, always rejects · BTC exempt
Rule 3 — ICT Force-Close Backstop: for every FILLED ICT signal, monitor computes force_close_at = MAX(killzone_end_of_fill_session, fill+60min) · closes with close_source='KILLZONE_EXPIRY' · TP/SL/manual close still fire first · BTC exempt
Rule 4 — Last-Hour No New Trades: 15:30–16:30 ET blocks all non-BTC strategies (ICT, AMD50, LIQVOL, MEANREV, ORB_5M, ORB_15M) · BTC exempt

Bitcoin (BTC) — 24/7/365 · v240.2

Fully exempt from Rules 1/2/3/4 — BTC is not bound by US session killzones (not an equity-index/futures concept)
No EOD hard close — BTC positions remain open past 21:30 UK
Weekend signals allowed — BTC is the only asset that can trade Sat/Sun
Fill checker runs 24/7 — AGENT02._fill_checker_loop no longer gated to Mon–Fri 07–22 UK (v240.2)

Automated Schedule (UK Time)

07:00 + 20:00 UK dailycalendar — To-Do & Agenda brief + economic calendar written to MACRO:CALENDAR:UPCOMING · injected into daily brief LLM
07:00 UK Mon–Fri — Daily morning brief with market narrative + upcoming event context
07:30 UK Mon–Fri — Life Sciences digest (toggle in Alerts & Schedule)
08:00 UK Mon–Fri — London Open ICT brief · signal generation pass
11:00 UK Mon–Fri — Mid-morning ICT update
13:00 UK Mon–Fri — Pre-NY ICT brief · signal generation pass
17:00 UK Frireview — Weekly trade review Telegram summary
21:00 UK Mon–Fripatterns — EOD summary + evening memory digest · weekly summary auto-written Fri
09:00 UK Sat–Sun — Weekend brief · BTC price/range/headlines + Oxford weather
20:00 UK Sun — Sunday evening setup brief · Friday close review + Monday ICT setups
21:00 UK daily — Evening memory digest (every day inc. weekend)
Background monitor: every 60s checks all OPEN signals vs live price · auto-closes on TP or SL hit (0.1% tolerance)
Scheduler v3 — memory-driven · reads schedule:alertsschedule every 55s · UI changes live within 1 min · weekend-aware

Signal Rules & Cadence

State machine: ALERTED → OPEN (▶ Enter) → CLOSED (✓ TP / ✕ SL)
Strategy tagsignal — strategy_tag column in setup_alerts · values: ICT · AMD50 · LIQVOL · MEANREV · used for filtering in Trade History + Performance
Create Signal: + Create Signal button on AMD50/LIQ+VOL/Mean Rev tabs → inserts into setup_alerts with correct strategy_tag → Telegram fires → monitor picks up
AMD50 auto-create: score ≥7/9 + R:R ≥1.5 → auto-inserts with strategy_tag=AMD50 without manual button press
Cadence limits: HTF min 4h · LTF min 1h · SCALP min 15min between signals per instrument per direction
Suppression: entering a signal auto-suppresses all other ALERTED signals on same asset+direction
Zone end warning: ⚠ late badge if signal generated within 10 min of kill zone close
Manual override: ▶ Enter · ✓ Take Profit · ✕ Stop Hit · ⊘ Suppress per signal row
Close source: AUTO (TP/SL monitor) or MANUAL (user action) — logged per trade with timestamp
Telegram alerts: strategy-aware label · 🟢 Entry hit · ✅ Take Profit · 🛑 Stop hit

Learning Feedback Loop — 4-week data gate before strategy bias recommendations

✅ Edge profilesignal — TRADING:EDGE_PROFILE best/worst KZ combos + min confluence threshold · injected into every ICT brief + Ask Anything · gated to 28 days data minimum
✅ Trade memorylearning — every close writes TRADING:KZ_PERF:{asset}:{kz} · TRADING:CONFLUENCE:scoreN · TRADING:TRADE:{id}
✅ Drawdown gatesignal — session P&L ≤-$100 → size warning · ≤-$200 → stop trading · injected into LLM
✅ Weekly summarypatterns — TRADING:WEEKLY:LATEST auto-written Friday EOD · includes R:R analysis, best/worst setups, session performance, confluence factor win rates
✅ R:R analysispatterns — TRADING:WEEKLY:RR avg R:R offered vs won/lost · best trade of week
✅ Session performancepatterns — TRADING:WEEKLY:SESSIONS avg P&L per trade per kill zone
✅ Setup analysislearning — TRADING:WEEKLY:SETUPS best and worst individual setup descriptions
✅ Confluence win rateslearning — TRADING:WEEKLY:CONFLUENCE EMA/VWAP/RSI/BOS/FVG/Liquidity/OB/MACD win rate when present
✅ Ask Anythingcontext — edge profile + KZ performance + today session P&L injected into every query
✅ AMD 50% trackingsignal — TRADING:AMD50:LATEST · auto-creates signal at score ≥7/9 · 4-week data gate · strategy_tag=AMD50 on all trades
✅ LIQ+VOL trackingsignal — TRADING:LIQVOL:LATEST · manual confirm required · 4-week data gate · strategy_tag=LIQVOL
✅ Mean Reversion trackingsignal — TRADING:MEANREV:LATEST · grade A+/B/C scoring · 4-week data gate · strategy_tag=MEANREV
✅ Per-strategy Performance — Performance tab has All · ICT · AMD50 · LIQ+VOL · Mean Rev pills · each calls /api/performance?strategy=X · isolated equity curve + ratios
✅ Trade History strategy filter — ICT · AMD50 · LIQ+VOL · Mean Rev filter pills in Trade History tab
⏳ Scalping data — 5-min candle entries to be incorporated when available · system ready to ingest

v3 Phase 1 — Trading Schema Upgrades v233+

New Columns on setup_alerts (17)

confidence_score INTEGER 0-10 — operator confidence at signal creation
confluence_score INTEGER 0-9 — composite confluence rating
confluence_components TEXT JSON — e.g. ["EMA_stack","FVG","OB","VWAP"]
volatility_regime TEXT — LOW / NORMAL / HIGH / EXTREME (ATR-based)
trend_regime TEXT — RANGE / WEAK_TREND / STRONG_TREND (ADX-based)
macro_event_proximity TEXT — NONE / SAME_DAY / NEXT_DAY / THIS_WEEK
macro_event_name TEXT — FOMC / NFP / CPI / etc.
day_of_week TEXT — MON / TUE / WED / THU / FRI / SAT / SUN
trade_mode TEXT — PAPER / LIVE / SIMULATED
bias_state TEXT — WITH_TREND / COUNTER_TREND / NO_BIAS
structure_state TEXT — BOS_UP / BOS_DOWN / RANGE / CHOCH_UP / CHOCH_DOWN
liquidity_condition TEXT — SWEPT / BUILDING / THIN / UNKNOWN
close_reason TEXT — TP_HIT / SL_HIT / MANUAL / EOD / TRAILING / TIME_EXIT
timeframe_htf TEXT — 4H / DAILY / WEEKLY
timeframe_entry TEXT — 5M / 15M / 1H
decision_rationale TEXT — structured decision reasoning
edge_bucket TEXT — strategy_asset_session key for edge lookup (post-28d)
mfe_r REAL — Maximum Favourable Excursion (R-units, updated live by monitor)
mae_r REAL — Maximum Adverse Excursion (R-units, updated live by monitor)

New Columns on trade_results (9)

pnl_r_multiple REAL — P&L in risk units (auto-computed on close)
mfe REAL — Maximum Favourable Excursion (R-units, Phase 2)
mae REAL — Maximum Adverse Excursion (R-units, Phase 2)
close_reason TEXT — mirrors setup_alerts.close_reason
execution_grade INTEGER 1-5 — post-trade grading
thesis_grade INTEGER 1-5 — was the setup thesis correct?
rule_following_grade INTEGER 1-5 — did you follow the rules?
management_grade INTEGER 1-5 — trade management quality (stops, targets, exits)
loss_category TEXT — loss attribution: BAD_THESIS / BAD_TIMING / BAD_SIZING / RULE_VIOLATION / MARKET_REGIME / UNLUCKY
trade_mode TEXT — PAPER / LIVE
slippage_estimate REAL — points of slippage

New Tables (5)

trade_overrides — logs manual interventions (suppress, stop adjust, force close)
strategy_changelog — strategy parameter versioning without sample contamination
decision_journal — cross-domain decision log with rationale + outcome tracking
comps — comparable transactions database (M&A, funding rounds, multiples)
deal_reviews — periodic deal review log with thesis drift tracking

RR Enforcement — 4 Layers (v232+)

✅ api_signal_create — blocks R:R < 1.5 on create
✅ AMD50 auto-create — blocks R:R < 1.5 before DB insert
✅ api_signal_suppress_check — rejects R:R < 1.5
✅ _monitor_open_signals — auto-suppresses ALERTED/PENDING with R:R < 1.5 every 60s
✅ AGENT02 _log_alert() — 1.5× RR gate + MGC $350 risk cap + BTC 3.85% stop cap + v3 field capture (v234)

Post-Close Enrichment

_enrich_trade_result(conn, alert_id, close_reason) — auto-computes R-multiple + sets close_reason
Called from: EOD close · live monitor TP/SL · manual close (PATCH /api/signals/state)
✅ v234: _backfill_trade_fields() retroactively computes pnl_r_multiple = pnl_dollars / 350 on all existing closed trades at startup

Decision Journal (v234)

GET/POST /api/decision-journal — list + create entries (domain, action, rationale, confidence)
PUT/DELETE /api/decision-journal/<id> — update outcome/retrospective, delete
Domains: TRADING · INVESTMENT · PERSONAL · STRATEGY
Outcomes: CORRECT · WRONG · MIXED · TBD
UI: Developer pill "Decision Journal" with domain filter, tabular log, edit/delete
_auto_journal() — auto-creates entries on signal create/close/suppress, TP/SL hit, EOD close, deal status change (v235)

Comps — Comparable Transactions (v235)

GET/POST /api/comps — list (optional ?deal_id= filter) + create
PATCH/DELETE /api/comps/<id> — update, delete
Fields: company_name, sector, sub_sector, transaction_type, acquirer, date, amount, pre_money, ev_revenue, ev_ebitda, revenue_at_deal, stage_at_deal, geography, source, notes, deal_id
UI: LS pill tab "Comps" with deal filter, sortable table, median stats, add/edit modal
One-pager: linked comps card with median EV/Rev & EV/EBITDA

Confluence Capture (v235)

PATCH /api/signals/<id>/confluence — update confluence_components + confluence_score
UI: 12-option checkbox editor on ALERTED/OPEN signal cards (FVG, VWAP, EMA21, EMA50, Order Block, BOS, CHoCH, Liquidity Sweep, MACD, RSI, SMT, Volume Imbalance)
Closed/suppressed signals show static tags (read-only)

📊 Market Analysis Tab — Tier 1 & Tier 2

Tier 1 — Live Data

ATR Context Bar: daily range used % vs ATR(14) per asset · price-in-range dot · green <50% used · red >80% (range exhausted)
Range Monitor: today's H-L vs 20-day average · EXPANDING / NORMAL / EXTENDED signal · best entries when <50% of avg range used
Signal Quality Tracker: win rate by asset + tier · last 30 days · populates from trade_results
Session P&L Heatmap: win rate per asset × kill zone · infers session from alert_time if session field empty

Tier 2 — Builds Over Time

Drawdown Monitor: session P&L · peak · current drawdown · equity curve SVG · ⚠️ caution at $100 · ⛔ stop at $200
Confluence Breakdown: parses signal rationale text for 8 factors (EMA/VWAP/RSI/MACD/BOS/FVG/Liquidity/OB) · score vs win rate chart · per-signal checklist
Kill Zone Performance: win rate + expectancy ($/trade) per asset per kill zone · 60-day lookback · green = trade it · red = skip
Liquidity Profile: average hourly volume from 30-day yfinance 1H data · kill zones highlighted · reveals true liquidity windows per instrument

✅ Memory integration complete — Every closed trade writes KZ performancesignal, confluence scoressignal and edge profilesignal to memory. Economic calendarcalendar written daily at 07:00. Weekly summarypatterns auto-writes Friday EOD with R:R, best/worst setups, session performance, confluence factor win rates. All injected into LLM on every signal generation and Ask Anything call.

📡 TradingView Webhook — Full Setup Guide

Step 1 — Webhook URL

Your webhook endpoint (click to copy):


              https://honeybees-mac-mini.tail56e8bc.ts.net/api/tv/webhook

Step 2 — In TradingView

Open a chart → click the Alarm clock icon → Create Alert
Set your condition (price crosses, indicator, Pine Script etc)
Under Notifications tab → enable Webhook URL
Paste the webhook URL above
Set the Message to the JSON format below

Step 3 — Message JSON Format

For Pine Script strategy alerts:

{
  "ticker": "{{ticker}}",
  "action": "{{strategy.order.action}}",
  "price": "{{close}}",
  "interval": "{{interval}}",
  "message": "{{strategy.order.comment}}"
}

For simple price/indicator alerts:

{
  "ticker": "{{ticker}}",
  "action": "alert",
  "price": "{{close}}",
  "interval": "{{interval}}",
  "message": "Your signal name here"
}

Notes

Alerts arrive at Trading Intelligence → TV Alerts tab in real time
Each alert is stored in the memory table as tv:alert:{timestamp}alert
A Telegram notification is sent for every webhook received
The webhook endpoint is PIN-exempt — TradingView can reach it without a session
Tailscale Funnel must be running on the Mac Mini for external webhooks to reach AGENT00
Test by sending a manual POST: curl -X POST https://honeybees-mac-mini.tail56e8bc.ts.net/api/tv/webhook -H "Content-Type: application/json" -d '{"ticker":"ES","action":"alert","price":"5000","interval":"1h","message":"Test"}'

How Memory Works

Storage: SQLite memory table · key / value / category / updated_at · persists across all restarts and deployments
Soul prompt: Hard-coded standing context in AGENT00_memory.py — professional background, BTC thesis, H100 Advisory context, trading principles · injected into every LLM call
Context injection: Soul + all memory entries prepended to every LLM call — Ask Anything, Multi-Asset Brief, Daily Brief, ICT analysis all receive full context
Ask Anything: Every Q&A via web or Telegram calls process_message(question, answer) which extracts facts and saves to memory · also logged to event_log
Evening digest: run_evening_digest() — summarises all memory entries · sends to Telegram · refreshes context

What Gets Written to Memory

Key

Description

Used as

web:last_brief

Multi-Asset Brief text + timestamp

context web:last_daily_brief

Daily Brief text + timestamp

context web:last_indicators

Full indicator snapshot · cached 5 min

cache CONTEXT:TODAY:exchanges

Last 5 Ask Anything Q&As today · conversation continuity

context CONTEXT:ROLLING:facts

Last 20 decisions/facts extracted from Q&A · 7-day rolling

context MACRO:CALENDAR:UPCOMING

High-impact econ events next 7 days · FOMC/NFP/CPI/ECB with trading notes · written 07:00 daily

calendar MACRO:CALENDAR:JSON

Structured JSON of upcoming events for machine reading

calendar schedule:alerts

Telegram scheduler config · read every 55s by run_scheduler() · UI changes live within 1 min

schedule TRADING:EOD:YYYY-MM-DD

Nightly EOD P&L summary per asset

review TRADING:TRADE:{id}

Individual trade · asset/tier/direction/KZ/result/P&L/confluence · written on every close

learning TRADING:KZ_PERF:{asset}:{kz}

Running win rate + P&L per asset per kill zone · updated on every close

signal TRADING:CONFLUENCE:score{n}

Win rate by confluence score (0–8) · updated on every close

signal TRADING:EDGE_PROFILE

Best/worst KZ combos + min confluence threshold · 4-week data gate · injected into ICT briefs

signal TRADING:WEEKLY:LATEST

Full weekly summary text · patterns, asset/session/tier/direction breakdown

patterns TRADING:WEEKLY:RR

Avg R:R offered vs won/lost · best trade of week

patterns TRADING:WEEKLY:SESSIONS

Avg P&L per trade per kill zone session

patterns TRADING:WEEKLY:SETUPS

Best and worst individual setup descriptions with full context

learning TRADING:WEEKLY:CONFLUENCE

EMA/VWAP/RSI/BOS/FVG/Liquidity/OB/MACD win rate when present in signal rationale

learning TRADING:AMD50:LATEST

AMD 50% scan results · dealing range · EQ level · manip wick · SMT · inversion FVG · score/9

signal TRADING:LIQVOL:LATEST

LIQ+VOL scan results · Asian/London range sweeps · SMT divergence · prev-day H/L raids

signal TRADING:MEANREV:LATEST

Mean Reversion scan · capitulation scoring · grade A+/B/C · streak · move ratio vs ATR

signal TRADING:ORB:LATEST

Opening Range Breakout scan · 5m + 15m variants · OR high/low · RSI14 momentum · 1H bias · volume surge · score/10 · deduped by asset+direction (higher score wins, 15m on tie) · suggestion-only (no auto-create) · 5m expires 10:35 EST, 15m expires 11:00 EST

signal tv:alert:{timestamp}

TradingView webhook alerts · real-time · available in TV Alerts tab

alert

Current Status & Remaining Gaps

✅ Trading Intelligence Bloomberg restructure (v246): Overview now leads with TWO Bloomberg-style tables (instruments + macro) in matching format · ▣ MARKETS: BTC/MGC/MCL/MNQ/MES with LAST · NET CHG · % CHG · DAY RANGE bar · ▣ MACRO: VIX/DXY/UST10/Nikkei 225/Shanghai Comp (v246 added N225 + SSEC) with same column set · Macro Calendar tab deleted, its contents (econ calendar + FJ news) moved onto Overview · new Trade Setups pill right of Overview holds Multi-Asset Brief + EOD Summary · snapshot fix: today's high/low/open computed from today's bars only
✅ AGENT02 MNQ multiplier + race guard (v245): MNQ in AGENT02_trading.py CONTRACT dict fixed 0.50 → 2.00 (v225 fix only ever landed in web.py — AGENT02 was the orphan for the entire period, every AGENT02-monitored MNQ trade was priced at ¼ of correct) · _check_fills_once now (a) filters SELECT with sa.signal_state NOT IN ('CLOSED','SUPPRESSED') and (b) re-reads state inside the write transaction before UPDATE — fixes the race where AGENT02 clobbered manual closes made during its 60s tick (root cause of #271/#272/#278 WIN→LOSS flip) · CRITICAL comment added warning that MNQ multiplier must stay in sync with AGENT00_web.py PNL_MULT · 3 corrupted MNQ ORB rows healed via v245_heal_mnq_orb.sql
✅ Result-sign authority (v244): api_signal_state CLOSED branch derives result from the sign of computed pnl — never trusts client · sigClose client-side recomputes actualResult before POST and toasts a warn on button/outcome mismatch · fixes the #275 WIN-on-negative-pnl class of bug · EOD between-levels branch also fixed (was hardcoded LOSS) · log warning on any client/server disagreement for audit
✅ BTC stale-alert sweep (v243): hourly sweep inside _monitor_open_signals suppresses BTC ALERTED/PENDING rows older than 24h with no trade_results row · closes the gap left by v240.2 BTC carve-out (weekday EOD skips BTC) · close_reason='BTC_STALE_SWEEP_24H' · auto-journaled
✅ Legacy $0 P&L backfill ran (v243): v241_change_b_pnl_backfill.sql executed on SIMBA · healed legacy WIN/LOSS rows where pnl_dollars=0 but close_price was populated · uses identical formulas to v237 server-side rescue
✅ Session rules enforced in both paths (v240.1): Rules 1/2/4 gate api_signal_create AND AGENT02._log_alert — TradingView bypass closed · helpers duplicated in both files with CRITICAL comment blocks
✅ BTC 24/7 carve-out (v240.2): BTC fully exempt from all four session rules · _fill_checker_loop runs 24/7 · BTC is the only asset that trades weekends
✅ ICT Rule 3 force-close monitor: for every FILLED ICT signal, checks force_close_at = max(killzone_end, fill+60min) every 60s · Telegram alert on killzone expiry · BTC exempt
✅ Trade history timestamp audit (v241): close timestamp placement now driven by close_source + target_hit/stop_hit · ✓ under TARGET, ✕ under STOP, ✎ under CLOSE for manual, 🏁 for EOD, ⏱ for killzone expiry
✅ Confluence tag badges on trade history (v241): top 3 components rendered as cyan badges with +N overflow counter
✅ AGENT02 close-path parity (v241.1): _check_fills_once target/stop hit blocks now persist setup_alerts.closed_at + close_source='AGENT02_TP/SL' + close_price — previously a bypass that left 10+ legacy rows orphaned
✅ Trade history delete button (v242): per-row ✕ DEL in the status column · confirms with destructive-action dialog · auto-triggers /api/eod/recalculate-memory after delete · api_signal_delete also clears decision_journal entries (closes v235 orphan gap)
✅ Ask Anything memory extraction: process_message() confirmed working · extracts decisions/views/people/tasks
✅ Contacts in soul prompt: All SQLite contacts injected into every LLM call
✅ Brief context in soul: Last Daily Brief + Multi-Asset Brief included in memory context
✅ Trading edge profile in signal generation: TRADING:EDGE_PROFILE injected into every ICT brief + Ask Anything — LLM sees kill zone win rates, confluence thresholds, drawdown status
✅ Trade memory writer: Every closed trade writes kill zone performance + confluence score + outcome to memory — both manual close and auto-close monitor
✅ Conversation continuity: CONTEXT:TODAY:exchanges (last 5 today) + CONTEXT:ROLLING:facts (last 20 decisions, 7-day window) — injected into every Ask Anything · visible in Memory → Conversation tab
✅ Weekly trading summarypatterns — Auto-writes Friday EOD · 5 keys: LATEST · RR · SESSIONS · SETUPS · CONFLUENCE · visible in Memory → Trading tab
✅ Economic calendarcalendar — MACRO:CALENDAR:UPCOMING written 07:00 daily · injected into daily brief LLM narrative
✅ Scheduler v3 fixed — dead v2 code removed · try/except hardened · tick logs every 55s in agent00.log
✅ AMD 50% scanner — /api/signals/amd50-scan · score ≥7/9 auto-creates signal · strategy_tag=AMD50 written to DB
✅ LIQ+VOL scanner — /api/signals/liqvol-scan · sweep detection · SMT divergence · manual confirm required
✅ Mean Reversion scanner — /api/signals/meanrev-scan · 7-variable scoring · grade A+/B/C · 20MA target
✅ ORB scanner (v238): suggestion-only · 5m/15m variants · deduped by asset+direction · staleness flags (5m expires 10:35 EST, 15m 11:00 EST)
✅ Strategy signal creation — /api/signals/create endpoint · stores strategy_tag · strategy-aware Telegram label · feeds into full TP/SL monitoring flow
✅ Per-strategy Performance — /api/performance?strategy= · All/ICT/AMD50/LIQVOL/MEANREV/ORB_5M/ORB_15M pills · isolated ratios + equity curve · ⚠ banner when <10 trades
✅ Trade History filtering — strategy filter pills in Trade History · coloured strategy badges on each row
✅ Command Surface on MKT Technicals (v246.13–v246.22) — 6 stages: Focus Watchlist, Regime Strip, Driver Matrix, Risk Rail, Cause & Effect, News Reaction Tape. All live with background workers and dependency chain.
✅ Server-authoritative P&L (v246.15) — single _compute_trade_pnl() helper, api_signal_state never trusts client. +$1,738.76 recovered across 9 stale rows.
✅ Backup rotation fix (v246.22) — SQLite backup API, exception-resilient scheduler, startup diagnostic flags stale rotation in boot log.
✅ Risk Analytics on Performance page (v246.23) — distribution moments, tail risk (VaR/ES parametric + historical), payoff asymmetry, serial dependence (ACF + Ljung-Box), regime breakdown, path dependency, rolling Sharpe 20/60/120. Equity curve compounds from $100k start with $/% toggle. 30-min cached worker.
✅ Neuro Investment Intelligence rebuild (v247) — deleted old Neuro panel, rebuilt as Bloomberg-style investment surface: violet-gradient hero with 4 KPI tiles + stat strip, macro thesis row (tailwinds + payer logic), 8-subsector grid ranked by revenue weight, headlines feed preserved.
☐ BLOTTER / SCANNERS / EDGE / CONTEXT workspace restructure — Bloomberg-style 4-workspace UX revamp consolidating 14 tabs into 4 (not yet started, design doc only)
☐ Startup desync audit — boot-time check for signal_state vs trade_results mismatches, verify _monitor_open_signals is actually running
☐ Today Alerts headline split — show N actionable · M suppressed instead of single count
☐ Dynamic sizing enforcement — wire sizing_mult from expectancy engine into AGENT02 _log_alert actual position size
☐ PM override telemetry view — audit trail panel for Rule 2 overrides
☐ AMD50 last-hour gate gap — auto-create path bypasses Rule 4, needs audit
☐ api_macro off request path — currently blocks request while yfinance runs, move to background worker
☐ Route AGENT02 target/stop + EOD monitor through _compute_trade_pnl — finish eliminating pnl math duplication
☐ BTC stale-alert sweep refinement — 24h window covers most cases but edge cases around weekend rollover need verification
☐ Unified opportunity ledger — cross-domain ranked queue (trading + deals + research)
☐ Investment Phase 1 — schema + deterministic decision engine from design doc

Memory Roadmap

✅ Ask Anything extraction: process_message() extracts facts/decisions to CONTEXT:ROLLING:facts
✅ Weekly summarypatterns — R:R · best/worst setups · session performance · confluence factor win rates · 5 memory keys written
✅ Economic calendarcalendar — MACRO:CALENDAR:UPCOMING written 07:00 daily · FOMC/NFP/CPI/ECB with pre-release trading notes · injected into LLM narrative
✅ Conversation continuitycontext — CONTEXT:TODAY:exchanges + CONTEXT:ROLLING:facts in every Ask Anything
✅ Briefs cachedcontext — Daily Brief + Multi-Asset Brief persist and load on demand
✅ Scheduler v3schedule — memory-driven · reads schedule:alerts every 55s · dead v2 code removed · error-resilient
✅ Strategy taggingsignal — strategy_tag column on all signals · ICT / AMD50 / LIQVOL / MEANREV · filterable in Trade History + Performance
✅ Per-strategy ratios — Performance /api/performance?strategy=X · isolated Sharpe/Sortino/Calmar/equity curve per strategy · 4 strategy pills in UI
☐ TradingView alerts → memory context injection for LLM awareness of recent signals
☐ Scalping data (5-min candles) — system ready · data feed to be connected
☐ Economic calendar: live scrape fallback if investing.com blocks · OECD/ECB as backup
✅ Indicator snapshots cached — Signal Monitor loads from memory cache first

✅ v257 — AGENT00.py Rebuilt

AGENT00.py fully rebuilt (v257) — 691 lines. Previous orchestrator was lost in v247 git incident; all Telegram alerts were silently dead since then
9 send-points in web.py all working — from AGENT00 import send_telegram now resolves to a real function
✅ LS digest fix — _send_telegram bug at L16755 fixed (was undefined since v248)
✅ Drawdown alerts active — $200 / $350 / $500 thresholds, checks every 2 min
✅ 10 scheduled tasks — morning brief, ICT briefs, EOD, weekly review, weekend brief
⚠ TV webhook still unauthenticated — see Security section

Bot Config

Bot: @agent00X_bot
Chat ID: 542250707 · single user · private
Token: stored in .env — never committed
Process: nohup python3 AGENT00.py >> agent00_bot.log 2>&1 &

Commands (v257)

/menu — inline keyboard main menu
/pnl — today’s P&L with per-trade breakdown
/signals — active signals (ALERTED/OPEN/FILLED)
/brief — morning brief (weather, BTC, agenda, signals)
/btc — Bitcoin price + 24h change
/weather — Oxford 3-day forecast
/todo — open tasks with priority
/memory — memory bank stats
/status — system health + 7-day trading
/ask [question] — AI via OpenRouter
/futures — live futures table
/macro — global macro dashboard
/news — top market headlines

Live Telegram Send-Points (v257)

✅ api_signal_create — manual signal creation via UI
✅ _monitor_open_signals — 60s daemon: TP/SL hits, KZ expiry, EOD close, EOD cleanup
✅ api_signal_state — manual close/state changes via UI
✅ api_tv_webhook — TradingView POST → Telegram forward (unauthenticated)
✅ api_amd50_scan — AMD50 auto-signal with Telegram (manual trigger only)
✅ api_ls_digest — Life Sciences digest (fixed v257)

Proactive Monitors

Drawdown monitor (every 2 min): alerts at $200 · $350 · $500 daily loss thresholds
Fill checker: 🟢 ENTRY HIT · ✅ TARGET HIT · 🛑 STOP HIT (via web.py _monitor_open_signals)
Deduplication via in-memory set — no duplicate alerts per event

🔴 v248 Audit — Critical Findings

PIN gate is currently BYPASSED: DASHBOARD_PIN env var is empty/missing in .env, which causes require_pin before_request hook to return immediately without auth check (if not DASHBOARD_PIN: return). Whole dashboard is open to anyone reaching :8080. Fix: add DASHBOARD_PIN=<8+ digits> to .env on Mac, restart, verify pin screen reappears
TradingView webhook has zero auth: /api/tv/webhook is exempt from PIN check by design (external service can't authenticate via cookie), but has no HMAC signature, no shared secret in URL, no source IP whitelist. Anyone with the public Tailscale URL can POST arbitrary payloads that get stored in DB and forwarded to your Telegram
No rate limiting: If PIN is enabled, brute-force protection is absent. Even with PIN on, a script could try 10000 PINs/sec until lucky
No restart endpoint protection: /api/restart is gated by PIN session but if PIN bypassed (current state), restart endpoint is open
Single point of failure: SIMBA SSD failure = full system + DB loss. Time Machine drive arriving today (pending Dev To-Do #3)

Security Model (Design Intent)

PIN Authentication: Designed: dashboard protected by PIN on every visit · session cookie · 8-hour expiry · /logout clears session — currently disabled (see audit)
Session: Flask session with auto-generated SECRET_KEY stored in .env · regenerated on first run
Network: Dashboard on LAN (192.168.0.135:8080) + Tailscale remote access · no public port forwarding
Secrets: All API keys + PIN + SECRET_KEY in .env · .gitignore excludes .env and agent00.db
Telegram: Single Chat ID whitelist · all messages checked against TELEGRAM_CHAT_ID before processing
GitHub: Private repo · no secrets ever committed · DB excluded
TradingView webhook: /api/tv/webhook exempt from PIN (external service)
LLM: OpenRouter — no raw user data stored · prompts contain no PII beyond user-saved memory

Current Status (post v248 audit)

🔴 PIN authentication BYPASSED — see audit findings above
🔴 TV webhook unauthenticated — see audit findings above
✅ Tailscale VPN for secure remote access
✅ .env excluded from git · secrets never committed (when present)
✅ DB excluded from git · untracked
⚠️ .env file only on Mac Mini — back up to password manager or iCloud (Dev To-Do #4)
⚠️ Restart endpoint unprotected — accessible to anyone who can reach the API
⚠️ Time Machine pending — drive arriving today (Dev To-Do #3)
☐ Rate-limiting PIN attempts to prevent brute force
☐ HTTPS via Tailscale for LAN access (currently HTTP)
☐ HMAC or shared secret on TV webhook (Dev To-Do #2)

⚠ Architectural Risks & Scaling Plan (v248 audit)

Items that won't hurt today but will compound over months/years if not addressed. Each has a Dev To-Do entry.

File size: AGENT00_web.py is 49k+ lines. Hard to navigate, slow to restart, any syntax error breaks everything. Plan: extract market data, BTC ETF worker, econ calendar, earnings, investor intelligence into separate modules
DB grows unbounded: No retention policy on event_log, news caches, old setup_alerts. SQLite handles >1GB fine but query plans degrade. Plan: nightly retention job, monthly VACUUM in restart.sh
No DB indices audited: Likely missing on setup_alerts(alert_time), trade_results(close_time), memory(category), event_log(ts). Slow queries surface only when DB is large. Plan: EXPLAIN QUERY PLAN audit, add covering indices
24 background daemon threads, no observability: If any worker leaks memory or hangs, you'd only notice when something breaks. Plan: /api/system/threads endpoint with last-tick timestamps
No log rotation: agent00_web.log + agent00.log grow forever. Plan: RotatingFileHandler with maxBytes=10MB, backupCount=5
Single SSD = total loss risk: SIMBA failure loses DB, .env, logs, backups. Time Machine drive arriving today (Dev To-Do #3). Plan: + iCloud sync of backups + offsite encrypted snapshot to S3
Refresh cascade on boot: 24 workers all hit external APIs in first 30s. If yfinance/coingecko is slow, cascading delays. Currently mitigated by time.sleep() staggers but not formalized. Plan: priority-based startup queue
No external uptime monitor: If Flask process dies at 3am you find out by missing alerts. Plan: Uptime Robot free tier, hit /api/health every 5min
No request rate limiting: If PIN ever bypassed (currently is), endpoints with DB writes are open for hammering. Plan: Flask-Limiter once PIN restored, or rely on Tailscale exclusively
Old dead code accumulating: dormant /api/econ/calendar v1 (~321 lines), legacy renderers, old pill stubs. Plan: per-version cleanup pass

Loading health stats…

Current Network Setup

Mac Mini: Ethernet → 192.168.0.135 · local access at :8080
PC: Ethernet → 192.168.0.xxx · deploy via git push --force
Tailscale VPN: ✅ Installed · Mac IP 100.96.97.21 · access from any device
Tailscale Funnel: ✅ Running · public URL: https://honeybees-mac-mini.tail56e8bc.ts.net
TradingView Webhook: https://honeybees-mac-mini.tail56e8bc.ts.net/api/tv/webhook
Router: Virgin Media Hub · admin at 192.168.0.1
☐ Static DHCP reservation for Mac Mini recommended

Remote Access — Tailscale (Recommended)

Zero-config encrypted mesh VPN · free tier covers this use case
Mac Mini gets a stable private IP e.g. 100.x.x.x
Access dashboard from anywhere: http://100.x.x.x:8080
Works through hotel WiFi, mobile hotspot, firewalls
No port forwarding · nothing exposed publicly
Install Mac Mini: brew install tailscale && tailscale up
Install laptop: download from tailscale.com · login same account

Other Options

Cloudflare Tunnel — encrypted tunnel → HTTPS URL like agent00.yourdomain.com · no open ports · more setup
Port forwarding — NOT recommended · exposes dashboard publicly with no auth
Static DHCP: reserve 192.168.0.135 for Mac Mini MAC address in Virgin Media router so IP never changes

📋 Network TODO

☐ Set static DHCP reservation for Mac Mini → 192.168.0.135 in Virgin Media router (192.168.0.1)
☐ Install Tailscale on Mac Mini: brew install tailscale && sudo tailscaled & tailscale up
☐ Install Tailscale on laptop — login same account → dashboard accessible from anywhere
☐ Test laptop WiFi access: disable AP Isolation in router if needed
☐ Consider adding basic auth to /api/system/restart endpoint

Completed ✅

✅ PIN authentication on web dashboard
✅ AGENT06 — Personal Finance (PFIN) with FX conversion, all asset types, AI analysis, CGT, EIS/SEIS, mortgage type
✅ Signal state machine — ALERTED→OPEN→CLOSED, Take Profit/Stop Hit auto-resolution
✅ Kill zones — signals blocked outside Pre-Market/London Open/NY Open · clock bar indicator
✅ Micro contract tickers — MGC/MES/MCL throughout all display + correct multipliers (MNQ $2/pt, MES $5/pt, MGC $10/oz, MCL $100/pt, BTC % on $10k)
✅ Market Analysis tab — Tier 1 (ATR, range monitor, signal quality, session heatmap)
✅ Market Analysis tab — Tier 2 (drawdown monitor, confluence breakdown, kill zone performance, liquidity profile)
✅ Ask Anything conversation thread with full history
✅ Developer live sections — agents, health, database, history
✅ DB hourly backup + header button + iCloud todo
✅ Full-width layout · orange action buttons · API usage + last git push in clock bar
✅ Tailscale VPN + Funnel · TradingView webhook
✅ Contacts hub · H100 NAV · Life Sciences · Economic calendar
✅ v223 — LIQ+VOL · Mean Reversion · AMD 50% strategy tabs · strategy tagging on all signals · per-strategy Performance filters · Personal Finance tab (DE/Amy/Overview) · pfin AI analysis · CGT PPR exemption logic · daily/multi-asset brief fix
✅ v224 — Signal hard cap (1 per asset) · auto P&L calculation from close price · correct micro multipliers (MNQ $2/pt · MCL $100/pt) · signed pts column · Manual Close column in history · LIQVOL scanner rebuilt · cancel-all-alerts button
✅ v225 — Risk cap bypass fix (AGENT02 _log_alert() direct DB path now capped) · BTC P&L sign + notional corrected (dynamic $350 ÷ stop%) · MNQ multiplier $2/pt fixed · EOD false positive UTC→UK time fix · duplicate signal deduplication · per-strategy independent signal cap · 4-week paper tracking gate (28-day EARLY SIGNAL warning)
✅ v226 — Investment Intelligence renamed from Life Sciences · Deal Pipeline OS (three-layer: pipeline home, one-pager, edit modal) · 30+ field schema (financials, management, risks, scenarios, market) · Import Deck feature (PDF/PPTX→OpenRouter extraction, only fills empty fields) · currency per deal + Local/USD view toggle · A4 PNG export · Memory pills restored (switchMemTab) · Developer pills restored (showDevSection) · EOD time guard (blocks before 20:00 UK) · EOD historical records corrected (23/27/30 Mar)
✅ v227 — Business Overview section (tagline vs substantive) · TAM/SAM/SOM inputs + pills · Pedigree above management avatars · comma-split names · GET /api/ls/deals as SELECT * · asset_value / current_cash / gross_margin fields · financial notes independent · deal terms deduplicated · _fmtCcy hardened vs numeric input
✅ v228 — SOTP valuation: (EBITDA × multiple) + Asset + Cash − Debt · exit year selector per scenario · KPI hero MOIC/IRR/ROI linked to live SOTP calc · dilution scenarios (100%/75-25/50-50) in modal + one-pager + PNG · PNG rebuilt with SOTP figures · duplicate returns card removed · fix: _opFmt before-define ReferenceError silently killed entire one-pager render
✅ v229 — 2-page PDF export (new window, full A4 HTML, page 3 disclaimer) · BTC P&L fixed: ×2000 → dynamic notional ($350÷stop%) across all 5 locations (sigClose, auto-close monitor, AMD50 cap, suppress-check, renderHistTable) · min 1.5× RR enforced server-side · RR colour-coded (red/amber/green) · 1-per-strategy-per-asset rule — ICT/AMD50/LIQVOL/MEANREV/ORB_5M/ORB_15M each independent, no stack within same strategy across HTF/LTF/SCALP · 8 historical trade P&Ls corrected in DB · BTC cap check uses stop% not ×2000
✅ v230 — Telegram todo/agenda fix (wrong columns) · 20:00 schedule added · logger defined in AGENT00.py · resilient dispatch wrapper
✅ v231 — MGC $/pt docs fix · risk cap alignment · EOD sort + recalculate endpoint · Daily P&L chart labels · $0 gap fill · Import Deck restored + 8192 tokens · HMA parent pill + Health/Tech SPAC pipelines (zero duplication) · LS Headlines feed (15 RSS, 5 neuro-dedicated, 80+ keywords, tag filters)
✅ v232 — Investor Intelligence: 3 DB tables (investors, deal_investors, investor_tags) · 12+ API endpoints · profile view + tags + deal linking · sector×stage matrix · co-investment map · multi-select Type+Tag filters · 🤖 AI Extract from Deals · 🔍 Research Sector (5-batch, 40/batch, geographic quotas, must-include names) · portfolio companies field · controlled tag vocabulary + normalization (40+ variations) · 1.5× RR enforced at 4 layers (create, AMD50, suppress-check, monitor backstop)
✅ v236 — v3 Phase 2 started: Expectancy engine (/api/trading/expectancy — per strategy × asset × session buckets with R-multiple expectancy, profit factor, confidence levels, dynamic sizing recommendations, edge decay detection) · Strategy Health Dashboard ("Edge" trading tab — heatmap, sizing legend, decay alerts, grand totals) · Edge gating in signal creation + suppress-check (auto-blocks signals from negative-expectancy buckets with ≥20 trades) · Kill zone × asset session heatmap · Confluence factor analysis (/api/trading/confluence-analysis — per-factor win rate, score breakdown, best factor pairs) · Auto-journal on edge gate blocks
✅ v237 — Regime-specific expectancy (VOL×TREND heatmap + per-asset drilldown) · Regime-aware edge gating (signal_create + suppress-check) · MFE/MAE live tracking (60s monitor, signal badges, Edge analytics with capture ratio) · Confluence fix (editor on closed trades, NLP auto-backfill, one-click endpoint) · Regime backfill from historical yfinance data · Post-trade diagnosis system (4-grade panel: thesis/entry/management/rules 1-5, loss attribution: BAD_THESIS/BAD_TIMING/BAD_SIZING/RULE_VIOLATION/MARKET_REGIME/UNLUCKY) · Strategy Health Report (/api/trading/strategy-health — per-strategy grades, loss breakdown, weekly trends, auto-recommendation: PROMOTE/STANDARD/DEMOTE/PAUSE/RE-TEST) · Edge decomposition on Edge tab · Unified 9-axis underwriting framework for deals (Market/Founder/Technical/Timing/Capital/Regulatory/Exit/Downside/Strategic) · Server-side P&L computation on trade close (never trusts client $0) · 60s live P&L refresh · EOD stale signal cleanup (auto + manual Clear Stale button, safe — never touches trades with results) · Telegram notifications on all close paths (manual, auto-TP/SL, EOD) with strategy tag + entry/close price · Market conditions bar recalibrated to 4H/1H only (ignores 15M noise) · Trade ID shown in history table · ORB strategy (Opening Range Breakout) — 5m + 15m variants in single ORB pill, MNQ/MES/MGC/MCL, scores 0-10, auto-creates at ≥7, ORB_5M / ORB_15M independent slots, full integration (Performance pills, Trade History filter, Edge tab, Strategy Health, TRADING:ORB:LATEST memory, confluence tags ORB_Break / ORB_Retest / ORB_Volume)
✅ v238 — ORB refactor — auto-create removed (scanner is now a pure suggestion engine, no DB inserts on scan, no Telegram on scan) · 5m/15m deduped by asset+direction (higher score wins, 15m on tie for cleaner signal) · staleness flag per variant (ORB_5M valid until 10:35 EST / 60min after 9:35 OR close, ORB_15M valid until 11:00 EST / 75min after 9:45 OR close) · expired cards show ⏱ EXPIRED badge and hide Create Signal button · TRADING:ORB:LATEST memory reflects deduped suggestion-only state
✅ v239 — Hero counter ghost-count fix — renderSignalTable now skips signal_state='SUPPRESSED' rows in the openCount/pendingCount/win/loss/byTier loop, fixing the "2 live · 1 awaiting entry" phantom display when no real active trades exist (root cause: counter loop only inspected trade_results.result_status, never checked setup_alerts.signal_state, so SUPPRESSED rows with stale OPEN/NULL result_status were counted as live)
✅ v240 — Kill Zones & Session Rules — 4 canonical ICT kill zones defined in ET (source of truth: ICT_KILLZONES_ET): London 02:00–05:00 · NY Open 08:30–11:00 · NY PM 14:00–15:00 (observe-only) · Asia 20:00–24:00 · Rule 1 — api_signal_create with strategy=ICT rejects entries outside all 4 kill zones (KILLZONE_GATE_BLOCKED) · Rule 2 — NY PM is observe-only by default; rejects with PM_OBSERVE_ONLY unless user passes pm_override=true in the POST body, override trades count as real and are journaled as PM_KILLZONE OVERRIDE · Rule 3 — ICT force-close in monitor loop: for FILLED ICT signals, computes force_close_at = MAX(killzone_end_of_fill_session, fill_time + 60min), closes via same path as EOD with close_source='KILLZONE_EXPIRY', TP/SL/manual close still fire first, Telegram sends ⏱ ICT Killzone Expiry notification with zone name · Rule 4 — Last-hour gate 15:30–16:30 EST rejects all non-BTC api_signal_create with LAST_HOUR_BLOCKED, BTC exempt · ET-based helpers (_now_et, _ict_killzone_for_et, _is_pm_killzone_et, _is_last_hour_et, _ict_force_close_at_et) avoid UK-DST ambiguity · all four gates auto-journal blocks for later audit
✅ v240.1 — AGENT02 gate parity — ported v240 Rules 1/2/4 into AGENT02_trading.py _log_alert() so TradingView-fed ICT signals are subject to the same discipline as manual web-UI entries (was previously a bypass: _log_alert does a direct INSERT INTO setup_alerts, completely skipping api_signal_create) · same killzone definitions, same last-hour window, same ET-based helpers (prefix _v240 to avoid namespace collision with existing AGENT02 helpers) · PM observe-only in AGENT02 ALWAYS rejects because there's no user-in-the-loop override channel for TradingView ingestion — manual override remains web-UI only · helpers are duplicated between AGENT00_web.py and AGENT02_trading.py — documented with CRITICAL comment blocks in both files warning that killzone changes must update both · Rule 3 (force-close) was already working for TradingView signals because it operates on setup_alerts regardless of insertion path · frontend error handling (PM_OBSERVE_ONLY, LAST_HOUR_BLOCKED, KILLZONE_GATE_BLOCKED) added to createStrategySignal and createAmd50Signal: PM offers purple override button with confirm dialog, last-hour/killzone show grey disabled blocked button
✅ v240.2 — BTC 24/7 carve-out — BTC is now fully exempt from all four session rules · api_signal_create skips Rule 1 (killzone) and Rule 2 (PM observe-only) for BTC · Rule 3 force-close monitor excludes BTC via AND sa.asset != 'BTC' · Rule 4 already exempted BTC · AGENT02_trading.py _enforce_session_rules_v240 returns BTC_EXEMPT early · _fill_checker_loop no longer gated to Mon-Fri 07-22 UK — runs 24/7 so BTC fills are checked on weekends and overnight (futures assets have stale yfinance feeds outside market hours so no spurious fills) · rationale: BTC is 24/7/365 and is not bound by US session killzones — the killzone discipline is an equity-index/futures concept, not a crypto concept
✅ v241 — Trade History audit fixes — Change A: sigClose fallback path no longer silently proceeds with undefined sig when cache+API both fail — now shows error toast and aborts, preventing the NaN-to-$0 LOSS row pattern that caused manual closes to occasionally vanish P&L · Change C: auto-close monitor now persists trade_results.target_hit / stop_hit on every TP/SL hit (was previously computed locally and discarded) · /api/signals/history SELECT extended to return those columns · zero ripple risk, confirmed via grep that nothing else currently queries them · Change D: renderHistTable close-timestamp placement rewritten — was gated by !close_price which post-v237 never fires because every close path writes close_price, so the ✓/✕ timestamps under TARGET/STOP had been silently suppressed on all modern trades · new logic drives placement by close_source + target_hit/stop_hit: AUTO+TP shows ✓ under TARGET, AUTO+SL shows ✕ under STOP, MANUAL shows ✎ under CLOSE, EOD shows 🏁, KILLZONE_EXPIRY shows ⏱ · legacy rows (pre-Change-C, target_hit/stop_hit null) fall back to result_status · timestamp prefers close_time_str, falls back to closed_at[11:16] · Change E: confluence_components JSON rendered as cyan tag badges on history rows — top 3 with "+N more" overflow counter · history table only, live signals table unchanged per spec · data was always in the payload, renderer just never used it · Change B (not a code edit): legacy $0 P&L backfill SQL provided separately, scoped to trades where close_price IS NOT NULL, uses identical formulas to v237 server-side rescue, 3-step workflow (preview → update → verify) for safety
✅ v241.1 — AGENT02 close-path parity fix — root cause of trades #120/142/164/166 showing no close timestamp: AGENT02_trading.py _check_fills_once target-hit and stop-hit blocks only wrote trade_results and completely bypassed setup_alerts, leaving closed_at, close_source, and close_price NULL on every TradingView-fed trade that closed via AGENT02 monitoring · same class of bypass as v240.1 (AGENT02 skipping web-UI discipline) · fix: both blocks now also UPDATE setup_alerts SET signal_state='CLOSED', closed_at, close_source='AGENT02_TP' or 'AGENT02_SL', close_price, close_time_str in the same transaction as the trade_results update · renderHistTable extended to recognize AGENT02_TP/SL as valid auto-fill sources (✓ under TARGET / ✕ under STOP) and to fall back to result_status when close_source is empty or AGENT02_LEGACY (defensive for backfilled legacy rows) · one-time DB backfill SQL provided separately for the 10 affected legacy rows (4 with NULL closed_at use tr.created_at, 6 with populated closed_at but missing close_source get close_source='AGENT02_LEGACY')
✅ v255–v256 — Systematic P&L OS, Performance Mega-Build, Equities Intelligence, Memory Automation — Performance — Systematic P&L pill: new [Systematic] pill (purple, right of Oil) shows all-assets performance with theoretical P&L (stop/target hit, no manual close) · backend: /api/performance?mode=theo replaces every pnl_dollars with computed stop/target P&L using asset multipliers (GC:10, CL:100, MNQ:2.0, ES:5.0, BTC:10000/ep) · _compute_risk_analytics() extended with pnl_mode param — when "theo", all metrics (equity curve, distribution, tail risk, payoff, regime, path dependency) use theoretical P&L · /api/risk-analytics?mode=theo bypasses cache for theo · frontend passes &mode=theo to both performance and risk-analytics APIs when window._perfMode==='theo' · banner: "SYSTEMATIC P&L — What if every trade hit its stop or target — no manual close" · Systematic Trade Log: full execution-history layout (13-column grid) below risk analytics · matches Execution History exactly: confluence score badges, factor tags (top 4 + overflow), strategy badges, rationale text, WIN/LOSS status · three P&L columns: Actual, Theo (with target hit/stop hit label), Diff · scrollable (400px max-height), filter pills by asset + strategy · hidden from PNG/PDF captures via #theo-trade-log exclusion · Compare pill: new [Compare] pill (cyan, right of Systematic) · /api/performance/compare computes both actual and theo stats for ALL + each asset · 13-column comparison table: Asset, Mode (Actual/System), Trades, W/L, Win%, P&L (with diff inline), Return%, Sharpe, Sortino, Max DD, Best Day, Worst Day, Avg/Day · dynamic insight box comparing systematic vs discretionary performance · Performance — Min/Max/Avg Daily P&L: new 3-card row on ALL performance pills: Best Day, Worst Day, Avg/Day · backend min_daily_pnl/max_daily_pnl added to /api/performance · Performance — Asset Filter Pills: pills moved ABOVE equity curve · only shown on All Assets + Systematic views (individual asset pages preserve original side-by-side layout) · clicking an asset pill now reloads risk analytics (equity curve + all panels) AND confluence for that specific asset · layout check uses _isSingleAsset flag to exclude THEO from single-asset rendering · Performance — Context Panels: dynamic context boxes added to P&L Distribution (assesses mean vs median, skew direction, kurtosis) and Path Dependency (current drawdown state, equity stability, recovery times) · Execution History — Theoretical P&L Column: muted second line under actual P&L showing theo +$350 (+$59) for every trade · WIN: (target-entry)×mult, LOSS: (stop-entry)×mult · Execution History — Event Day Tags: trades on major US economic release days (CPI, PPI, PMI, FOMC, GDP, NFP, ISM) get red EVENT badge below date · uses cached Forex Factory calendar data · Overview — ETF/Futures Dedup: 20+ ETF→futures mapping (GLD→GC, USO→CL, QQQ→MNQ, SPY→ES, IBIT→BTC) · if futures shown in hero movers, corresponding ETF skipped · Overview — Topix Futures Deleted from backend fetch + frontend ORDER · Overview — VIX/UST10Y Net Change: % Chg, MTD, YTD, 12M columns now show net point changes instead of dashes (VIX/yields already in %) · BTC Treasury News Expanded: 4 new RSS feeds (CoinDesk, Forbes Crypto, Seeking Alpha, Reuters Business) · yfinance headlines per ticker 10→20 · cutoff window 7d→14d · H100 confirmed in companies list · Memory — Auto EOD Scheduler: new _eod_scheduler_loop daemon thread runs at 22:00 UK weekdays · auto-triggers run_eod_summary() writing TRADING:EOD:* and TRADING:PERFORMANCE:* · auto-weekly summary on Fridays · Memory — 30-Day Asset Feedback: POST /api/eod/write-performance-feedback standalone endpoint · writes directly via SQL (bypasses AGENT02 import chain) with per-asset stats, tier/direction breakdown, learning instructions · removed destructive 28-day gate that was deleting entries (used signal_state='CLOSED' which returned no rows) · Memory — Weekly Summary Fix: all_factors referenced-before-assignment bug fixed (moved definition before summary dict) · endpoint made synchronous for debugging with error propagation · Equities Intelligence — Indices Top 10 pill (default view): NASDAQ 100 top 10 by weight (AAPL 8.9%, MSFT 8.1%, NVDA 7.5%...) + S&P 500 top 10 by weight · columns: Name, Sector, Weight, Price, %1D, Mkt Cap, %MTD, %YTD, %12M · dedicated news feed with BTC-treasury-style ticker filter pills · Equities — Watchlist Enhancements: P/E ratio + Market Cap columns added (uses t.fast_info not slow t.info) · ticker filter pills above table · news feed below with ticker filters · Equities — Leopold Portfolio pill: full Q4 2025 13F holdings for Situational Awareness LP (28 positions, AUM $9.3B, AI/AGI focus) · columns: Holding, Type (COM/CALL/PUT), Price, %1D, Shares, Value($K), Weight, P/E, %MTD, %YTD, %12M · options color-coded (CALL=green, PUT=red) · dedicated news feed with ticker filters · Unified News Filter Design: _renderEqNews(el, articles, countId, filterId) reusable function · BTC-treasury-style filter pills: [ALL 45] [AAPL 8] [MSFT 6]... · orange active state with colored background/border · sort by pub_ts descending on both initial render and filter clicks · /api/yf-news/{ticker} endpoint: 10 most recent headlines per ticker via yfinance · PDF Standardization: _capturePagePDF(containerId, title, filename) reusable function · captures original DOM (not clone), hides .png-dl/.pdf-dl during capture · branded header with orange separator · "Private and Confidential property of Donald Ewer" footer on EVERY page · PNG/PDF Icon Standardization: 28 PNG buttons + 9 PDF buttons standardized across all pages · .png-dl 20×20 white SVG download arrow · .pdf-dl 20×20 white SVG document+arrow · Trading Rules v255: Signal creation rules (C: no dup today, D: single best per asset, E: no opposite to open, F: skip if open today, G: HTF blocks LTF/SCALP, H: max 2 pending/day) · Fill checker rules (A: same asset open→manual, A2: opposite dir→warn, A3: HTF open+LTF/SCALP→manual, B: score<5→manual) · Asia kill zone changed to observe-only · Critical Bug Fixes: nested backtick in template literal (EVENT tag), duplicate const r12Up (VIX/UST10 code), trade log in wrong function scope (crashed all performance pills), all_factors before assignment (weekly summary), tr.close_source column doesn't exist, 28-day gate deleting performance feedback, t.info blocking (replaced with t.fast_info) · 55,100+ lines
✅ v250–v251 — BTC Options OS: Full rebuild with Deribit live API + Treasury Risk Management System — Deribit API integration: _fetch_deribit_live() REST calls to GET /api/v2/public/get_index_price + get_book_summary_by_currency for live spot, mark_iv, OI on all BTC options · 60s cache · simulated fallback if Deribit unreachable · symmetric strike grid (5+ATM+5, 65–140% of spot) · tenor selection spread across full curve (1W to 1Y via target_days matching) · interpolation for missing strikes · data source badge (green DERIBIT LIVE / amber SIMULATED) · 6-pill architecture: Overview · Surface Lab · Screener · Treasury Risk · History · BIG · Overview: 8-tile hero + 6-tile KPI row · regime engine (6 states: CARRY RICH, EVENT RISK, CRASH HEDGE BID, BULL CALL SKEW, COMPRESSED, DISLOCATED WINGS) · market context with gamma/OI commentary · IV surface heatmap (8-stop cinema-grade gradient) · 25d/10d skew tables · ATM term structure chart with contango/backwardation assessment · 7 strategy cards · PDF + PNG exports · Surface Lab: full surface explorer (420px canvas with Retina scaling) · smile slices with multi-select tenor toggles · forward vol ladder with context notes · OI concentration (horizontal stacked bars, call/put split) · GEX by strike (BS gamma computation, gamma flip detection) · PNG export · Treasury Risk (institutional-grade): options_positions DB table with entry Greeks · position entry form with strategy tags, moneyness quick buttons (OTM/ATM/ITM), auto-pricing from live BS/IV, premium % display, trade date/time, expiry dropdown with standard + Deribit tenors · position cards with 6-column financials (premium, P&L $/%, daily theta, return on capital, collateral in USD + BTC, break-even), entry vs current Greeks side-by-side, risk tags, BTC/sats return, attribution · Hero (3 rows × 5): Row 1 = Holdings/NAV/Premium/P&L/Eff.Strikes · Row 2 = Covered%+BTC+sats/VaR BTC-only/VaR Net/Options VaR Impact/Max Loss 99% · Row 3 = Net Delta BTC/Options Delta BTC/Gamma $/1%/Vega $/1vol/Theta $/d · Portfolio Greeks: 1st order (delta, gamma, theta, vega) + 2nd order (vanna, charm, vomma, speed) + coverage · Payoff diagram: 420px canvas, orange portfolio line, grey dashed BTC-only, green fill above zero / red below, SPOT marker · VaR: parametric VaR (1d/10d × 95%/99%) for BTC-only, net portfolio, and options impact % · Expected Shortfall (CVaR) 95%/99% with ES/VaR ratios · historical VaR chart (180d rolling 30d) with BTC price overlay · Theta P&L runway: projected cumulative income at 1d–90d forward · DTE traffic lights: critical ≤7d (red) / warning 7–30d (amber) / safe >30d (green) · What-If simulator: preview trade impact on delta/gamma/theta/vega/covered% without committing · Scenario NAV cube: 11 spot shocks × 7 vol shocks with $K P&L · Greek stress: 5 scenarios (BTC ±25%/10%/flat) · Collateral stress runway: margin % under spot-down/vol-up · Assignment ladder: probability per short position from BS d2 · Governance concentration: overwrite%/put-secured% vs limits, expiry + top-position concentration · Greek buckets by tenor: delta/gamma/theta/vega/vanna/charm by DTE bucket · Snapshot Greeks button: overwrites entry Greeks with current live surface · BIG pill: Metaplanet BFC 2026 deck data (40,177 BTC treasury, ¥8.58B BIG rev, capital instruments, roadmap) · Personal Finance: bank column + custom tags + debit/credit formatting · Design constants updated: section headings 13px/700/WHITE, KPI values 18–20px/800/mono, table headers 11px/#fafafa, body 12–13px, all numbers max 2dp with commas, $ signs, +/- prefix, canvas devicePixelRatio scaling · 44,600+ lines
✅ v249.20 — Google News merged into BTC TC feed + ETF holders fallback + wealth history FX fix + BTC holdings update — Google News consolidation: removed separate Google News card from BTC TC tab, merged into existing treasury company news feed · loadBtcTcNews() now fetches /api/btc/treasury-news AND /api/btc/google-news in parallel, deduplicates by title prefix, re-sorts by date · G·NEWS filter pill added to unified feed · ETF holders: added mutualfund_holders fallback when institutional_holders returns empty for ETFs · Wealth history FX corruption fix: root cause — _fetch_fx_rates_for_snapshot() returned {"GBP":1.0} on internal Flask call failure, treating SEK 4.4M Björnrike valuation as £4.4M GBP · three-layer fix: (1) hardcoded fallback FX rates (SEK=13.5, USD=1.27, CHF=1.12 etc), (2) sanity check rejects >50% day-over-day swings on save, (3) boot-time auto-purge deletes snapshots >2× minimum net_worth · BTC holdings update: hardcoded fallbacks updated to April 2026 values (MSTR 780,897, XXI 43,514, Metaplanet 40,177) · treasury-holdings endpoint converted from per-request fetch to 30-min cached background worker from bitcointreasuries.net (API first, HTML scrape fallback)
✅ v249.19 — EDGAR removed + panel reorganization + institutional holders — SEC EDGAR deleted entirely: backend (cache, worker, CIK map, endpoint) + frontend (JS loader/renderer/pills) — SEC blocked/rate-limited, dead code · Panel moves: Google News card Overview→BTC TC tab, Asia News card Overview→BTC Asia tab (replaces placeholder) · NEW: Top 10 Institutional Holders via yf.Ticker.institutional_holders (13F-sourced via Yahoo Finance, no API key) · GET /api/btc/holders with 60-min TTL cache, ThreadPoolExecutor parallel fetch for 6 BTC TC companies + 5 ETFs · returns holder name, shares, % outstanding, value, date reported · two cards: BTC TC tab (companies) + ETF tab (ETFs) with ticker pills + grid renderer
✅ v249.18 — Sweden P&L model correction + SWC.L GBX fix — FUNDAMENTAL P&L FIX: the comm field IS net revenue received (advert minus platform fee), NOT a cost · previous code subtracted it, inverting the entire P&L · column renamed "Less Comm" → "Net Revenue" · netProfit = netRevReceived − totalCosts · hero restructured: 6 cards (Gross Advert → NET REVENUE → Bokavem → Op Costs → Tax → Net Profit) · stats strip added: HIGHEST WEEK / AVG PER WEEK / LOWEST WEEK with dates · auto-calc ⚡ fixed: bokavem = advert×25%, comm = advert − bokavem · waterfall: Gross Advert → −Bokavem → =NET REVENUE → −Costs → −Tax → =NET PROFIT · SWC.L GBX fix: LSE/AQSE stocks trade in pence not pounds, price divided by 100 for mkt cap, displays as "36.85p"
✅ v249.17 — Sweden audit fixes + breakout 1D — BUG 1 (CRITICAL): bookedRevenue=0 when no BOOKED status — now treats all advert>0 rows as booked when no explicit status · BUG 2: comm/bokavem scope inconsistency fixed · deleted Per Wk £ / Per Person £ columns · revenue=green, costs=red with − prefix, GBP sub-lines on all hero cards + totals + waterfall · tax changed from violet to red · Breakout 1D: added prev_high/prev_low from daily bars to backend + frontend TFS array (['1D','4H','1H','15M'])
✅ v249.1 — Sweden tab rebuilt as multi-year ledger from real spreadsheet data — replaced the placeholder 3-tier seasonal calculator with a full multi-year editable ledger seeded from the user's actual Börnrike spreadsheet (2021/22 cost-only, 2022/23 + 2023/24 with full booking + cost data) · Year switcher pills: top-right pill row shows all seasons with inline net-profit tag (+/− SEK k), click to switch active year, + Add Year prompts for new label and seeds an empty 22-week template · Booking grid: 14-column editable table per ski-season week (50−52, 1−17 = ~22 weeks Dec→Apr) with WEEK / PEAK toggle / STATUS cycle (BOOKED / EWER / empty) / DAY / START / END / DAYS / ADVERT (SEK) / LESS COMM / BOKAVEM / PEOPLE / PER WK £ / PER PERSON £ / NOTE · row background tints lavender if booked, faint pink if peak · ⚡ button per row auto-computes commission as 25% of (advert − bokavem) · Cost ledger: 11 fixed line items with vendors (Fyrfasen AB / Härjedalen Ko / Björnrike SAM / IKEA / LF Jämtland / Skatteverket etc) · per-row SEK input + auto-converted GBP · tax row highlighted violet · KPI strip (8 cards): Gross Advert / Booked Revenue / − Commission / − Bokavem / Net Revenue / − Op Costs / − Tax / NET PROFIT (color-graded green/red, GBP underneath SEK) · Year-over-year bar chart: 3 bars per year (gross green / costs red / net green-or-red), click any year to switch active · Property & FX panel: editable SEK→GBP rate, default commission %, area m², rooms, beds · avg per booked week + avg per person callouts · P&L waterfall: gross → commission → bokavem → net rev → op costs → tax → NET PROFIT line by line in SEK + GBP · State management: window._swedenModel deep-cloned from defaults so edits don't pollute the seed; year add/delete/reset operations all wired · NOT changed in v249.1: still no DB persistence (POST /api/pfin/sweden snapshot endpoint queued for v250); user can fill in 2023/24 actuals as season unfolds · previous simplified Sweden 30%-flat tax calc removed in favour of per-year actual tax line item from Skatteverket
✅ v249 — Sweden tab + BTC Options/Credit placeholders + manual entry-price + signal score columns — Personal Finance → Sweden tab: new pill right of Amy hosting a full Börnrike ski-house P&L calculator · editable inputs for 3-tier seasonal weekly rental rates × weeks (high/mid/low season), mortgage balance & rate (interest auto-computed), utilities/insurance/HOA/maintenance, cleaning per turnover (auto-multiplied by total weeks), agency mgmt fee % · simplified Sweden non-resident tax: 30% flat on (gross − SEK 40k standard deduction − 20% of gross) · SEK→GBP conversion editable · live KPI strip (gross / total costs / tax / net post-tax) and full waterfall P&L summary panel · in-memory state via window._swedenModel — no DB persistence yet, snapshot endpoint deferred · showTab('sweden') hides standard pfin-hero/sections/add-btns and reveals dedicated Sweden panel · BTC Intelligence → Options + Credit pills: two new placeholder tabs right of BTC Asia matching the Bloomberg visual language · Options stub: IV surface / block flow / structured plays cards · Credit stub: convertible tracker / collateralised lending / mNAV premium cards · both wired into showBtcTab as no-op tabs ready for content · Manual entry-price correction: ICT signals fire with fixed entry/stop/target but ORB/AMD50/LIQVOL/MEANREV require manual confirmation and actual fills can differ from suggested entry by ticks/points · new PATCH /api/signals/<id>/entry-price endpoint accepts {entry_price: float} · v246.15 lock honoured: refuses with HTTP 409 if signal_state=CLOSED or result_status in (WIN,LOSS) since server-authoritative P&L would be silently invalidated · recomputes risk_reward from new entry vs existing stop/target · writes decision_journal entry tagged ENTRY_PRICE_CORRECTION with old→new delta and new R:R for audit trail · frontend: small ✓ button on Open Trades Entry cell, gated to non-ICT & non-closed only · editSigEntry() handler prompts for fill price, warns if >5% deviation, toasts result · Signal score column: new dedicated column in both Open Trades grid (12→13 cols) and Trade History grid (12→13 cols) · reads confluence_score (0-9) already populated by ICT generator · color graded green ≥6 / amber 3-5 / red 1-2 / dash if 0 · history SELECT extended with sa.confluence_score · NOT changed in v249: Telegram code still pending (deferred from v248 audit), TV webhook still unauthed, MFE/MAE backfill still queued, no DB persistence on Sweden inputs, no real content in Options/Credit placeholders
✅ v248.1 — Security restoration + repo sync + handoff — Critical security fixes: DASHBOARD_PIN restored in .env (was empty since recovery, dashboard had been wide open) · SECRET_KEY set to fresh 64-char hex value via secrets.token_hex(32) · PIN screen verified working post-restart · Repo sync incident: Mac was 95 commits ahead of GitHub for an unknown duration (push hadn't happened in days) · pushed all 95 commits in one batch including the AGENT00.py recovery + v247.19* + v248.0 changes · PC sync: git pull on PC also deleted a stale 163KB agent00.db stub that had been tracked since project setup, replaced with proper untracked behavior · Git history privacy finding: agent00.db had been accidentally committed in 3 commits historically (project init Mar 11, agent rename Mar 14, v237 release Apr 6) · private repo so impact limited but file is permanently in history · .gitignore now hardened against agent00.db / *.db-* / *.bak_* patterns to prevent recurrence · Backup files moved: AGENT00.py.bak_v247.19d_pre_dbfix + AGENT00.py.bak_v247.19e_pre_ff + patch_agent00_ff_econ.sh relocated from working tree to /backups/recovery/ · MFE/MAE diagnosis: investigated empty MFE/MAE columns — mechanical limitation, not a bug — the _monitor_open_signals 60s tick rate combined with yfinance 15-min delay means trades closed within 5-30 minutes (typical manual-close pattern) get sampled at most 5 times with stale prices — need realtime data feed or post-close backfill from 1-min OHLCV bars to fix properly (Option 3 backfill in _enrich_trade_result is cheapest, queued for v249) · Decision Journal usage gap: auto-journaling firing correctly on every signal create/close/suppress/gate-block but the human-in-the-loop fields (confidence, outcome, retrospective) are empty across all entries — system half is working, behavioral half (Friday review ritual) needed to unlock value · Developer section: page-sub bumped v248 → v248.1, history entry for this work added · Handoff document: AGENT00_HANDOFF_v243_to_v248.md written covering the full arc since the v246.22 handoff (BTC sweep, command surface, Neuro rebuild, earnings, econ calendar, BTC Asia, AGENT00.py disaster + recovery, security audit) · NOT changed in v248.1: Telegram code (still pending v249 cleanup), TV webhook auth (still no HMAC), MFE/MAE backfill (deferred to v249), all background workers
✅ v248 — System audit + econ calendar refinement + Developer overhaul — Econ calendar (Forex Factory): filter narrowed to impact=High (red folder) + central-bank events only (FOMC, ECB, BOE, BOJ, PBOC, rate decision, press conference, meeting minutes, monetary policy, MPC) via new _CENTRAL_BANK_PATTERNS tuple in _refresh_ff_econ_cache · drops the medium-impact noise (PMI revisions, housing surveys, sentiment) across all 5 regions · expect 8–15 events/week vs prior 50+ · sort changed to today-first then chronological via (0 if e["is_today"] else 1, e["datetime_uk"]) tuple key so today's releases group at top of card regardless of time-of-day · Audit findings logged in Dev To-Do: (1) DASHBOARD_PIN empty in env so PIN gate is bypassed — entire app open to anyone reaching :8080, urgent restore needed · (2) TradingView webhook /api/tv/webhook has zero auth — anyone with URL can POST arbitrary alerts that Telegram-forward and store in DB · (3) Telegram audit: 8 send-points mapped (TV webhook, signal create, AMD50 scan endpoint, LS digest, _monitor_open_signals 4×, signal state) — _send_telegram at L12522 is undefined (LS digest crashes), AMD50 scanner endpoint exists but isn't scheduled, no drawdown alert wired · (4) Architectural risks: AGENT00_web.py 36k lines becoming unwieldy, no DB retention/archive policy, no index audit, 24 background threads with no observability, no log rotation, single SSD failure = total loss until Time Machine drive arrives today · Developer section refresh: page-sub bumped from v224 → v248, Dev To-Do list expanded from 15 to 25 items adding security findings, Time Machine pending entry, Telegram cleanup task, architecture/retention/index/log-rotation TODOs · NOT changed in v248: Telegram code (audit only, cleanup deferred to v249), PIN config (user must set in .env), all background workers untouched · logger now reports "cached N High+CB events" to make filter visible in agent00_web.log
✅ v247.19e — Forex Factory econ calendar + earnings released split — replaced procedural day-of-month econ calendar generator with real data from Forex Factory's free weekly JSON feed (nfs.faireconomy.media/ff_calendar_thisweek.json) · new _FF_ECON_CACHE + 30-min background worker (rate limit 2 req/5min, our cadence is 1/30min, well under) · _refresh_ff_econ_cache handles HTML "Request Denied" rate-limit response gracefully (keeps stale cache) · _parse_ff_datetime_to_uk converts ET-with-offset to UK local naive via zoneinfo (pytz fallback) · new endpoint /api/econ/calendar/v2, old /api/econ/calendar kept dormant as one-release safety fallback · frontend rewrite: 7-column grid (impact dot/date/time/event/previous/forecast/actual), today-in-red rule using server-computed is_today flag, beat/miss badges via _econBeatMiss numeric parse, USA default region pill · earnings card split into Pending (no eps_actual) + Released (actual populated, muted opacity 0.72) subsections with section headers and counts · Mac-side patch script patch_agent00_ff_econ.sh upgrades AGENT00.py _write_econ_calendar_to_memory to use same FF source, drops broken investing.com scrape and stale KNOWN_EVENTS array
✅ v247.19d — BTC Asia tab + AGENT00.py recovery + _db fix — new BTC Asia pill in Bitcoin Intelligence tab row (right of ETF) · intelligence hub for Philippines BTC treasury company advisory deal · hero with 6 macro stats: Metaplanet (40,177 BTC, 3rd globally) / PH crypto adoption (#8 globally, 11M users) / remittance TAM ($38.3B/yr) / crypto txn volume (~$40B/yr) / Strategic Reserve Bill HB 421 (10k BTC) / CASP framework (₱100M min) · comparable treasuries table (Strategy/Twenty One/Metaplanet/PH BTC Co TBD) · 6 content sections stubbed: thesis, deal structure (4 vehicle candidates), regulatory environment, investor target profile, research vault, news feed · investor cross-linking deferred per user instruction · Disaster recovery: AGENT00.py was silently deleted in v246.23 Risk Analytics commit dc7ce7f (2,202 lines lost); recovered from git blob 5cd7f43c via git cat-file -p after grep through git rev-list --all --objects · launchd plist patched with -u flag for unbuffered Python stdout so scheduler tick logs become visible in agent00.log · _write_econ_calendar_to_memory bug fixed: function used with _db() as conn: but bot module had no top-level _db helper — injected local sqlite3 + contextlib context manager + replaced 2 latent datetime.now() calls with _dt.now() alias · verified via direct function call test, function now writes [CALENDAR] Written N events to memory cleanly

Neuro Investment Intelligence page rebuild

Hero section

Macro Thesis row

Subsectors grid

Headlines feed preserved

id="ls-news-neuro"

loadLsPillarNews('neuro')

Design discipline

#a78bfa

--surface-2

--border

#fafafa

var(--mono)

#c9c9d1

Integrity verified

node --check

id="lssec-neuro"

id="ls-news-neuro"

showLsSection('neuro')

✅ v246.23 — Risk Analytics module on Performance page — new full daily-P&L statistics suite rendered as 8 stacked panels beneath the Drawdown from Peak chart · Backend: new _compute_risk_analytics() pure function reads setup_alerts ⋈ trade_results, rolls up daily P&L, carry-forwards zero returns on trade-less weekdays (industry standard for time-series stats), compounds equity from fixed $100,000 starting NAV using prior-day EOD → next-day SOD (industry-standard time-weighted approach) · new _RISK_ANALYTICS_CACHE + 30-min worker with belt-and-braces exception handling and tick logging (v246.22 daemon-death lesson) · new _risk_sanitize() recursive NaN/Inf scrubber so jsonify never chokes · new /api/risk-analytics endpoint, cache-first for default ALL/$100k, on-demand compute for filtered queries · new market_events table (date PK, event_type, severity) for manual stress-day tagging (CPI/FOMC/NFP/BTC_CRASH) · Metrics: distribution moments (mean/median/stdev/min/max + Fisher-Pearson bias-corrected skew + excess kurtosis, pure-Python no scipy dep), 20-bin histogram, parametric VaR/ES at 95/99 (normal assumption, z-scores 1.6449/2.3263), historical VaR/ES at 95/99 (linear-interp percentile, empirical tail mean), per-day payoff asymmetry (hit rate, payoff ratio, profit factor, expectancy/day, avg win/loss day, best/worst day), serial dependence (ACF lag 1-10 on raw + absolute returns for volatility clustering, Ljung-Box Q at lag 5 and 10 with χ² critical values, 95% confidence band), regime breakdown (volatility HIGH/NORMAL/LOW × trend UP/DOWN/FLAT + own-P&L sign + stress days), path dependency (time underwater, avg/max recovery days, pct days within 5/10% of peak, completed drawdown streaks table), rolling Sharpe (20/60/120-day windows, annualised) · Frontend: new loadRiskAnalytics(strategy) fires from loadPerfStats _drawWhenReady after base charts settle · 8 panels rendered into <div id="risk-analytics-container"> placeholder injected beneath drawdown chart · equity curve with $ / % toggle via setRiskEquityBasis() (redraws only the equity canvas, preserves other chart state) · 4 Chart.js charts (equity line, histogram bars, ACF grouped bars, rolling Sharpe triple-line with right-edge alignment padding for shorter windows) · insufficient-data gate: n_days < 20 shows amber banner + suppresses distribution/tail/payoff/serial panels but still renders equity curve + path dependency + regime breakdown · Data model decisions: starting capital $100k fixed inception (not lookback reset), compounding NAV (prior day EOD = today SOD), carry-forward zero days on weekdays only (weekends excluded unless existing BTC trade on that date), min 20 days for distribution stats / 20/60/120 for rolling Sharpe windows respectively · verified end-to-end against 50 synthetic trades / 29 days mock DB: all 7 metric families populated with sensible values, equity $105.9k (+5.9%), hit 62%, payoff 2.33
✅ v246.22 — Backup rotation fix — root cause: shutil.copy2 on live SQLite in WAL mode was producing torn writes AND the scheduler thread was dying silently on the first exception (daemon threads die with no log) · on 10 Apr around 12:36 the thread died after one tick leaving 20 "hourly" backups all with identical mtime from a single batch operation · Fix 1: _backup_db rewritten to use SQLite native backup API (src_conn.backup(dst_conn)) instead of shutil — safe for live DB with WAL/SHM sidecars, returns True/False instead of None so caller can detect failures · Fix 2: _backup_scheduler wrapped in belt-and-braces try/except so the daemon thread cannot die — every tick logs success or skip with a tick counter · Fix 3: new _backup_startup_diagnostic() runs at boot and logs the newest backup's age — if >4h old it logs ERROR with "STALE, rotation was broken" marker so any future failure is visible in the first seconds of the startup log · Fix 4: de-dup check uses glob with current-hour prefix instead of reading latest filename, more robust to filename parsing edge cases · Fix 5: backup file size verified >1KB before logging success, catches empty/truncated writes · boot sequence reordered: diagnostic → scheduler thread start → initial forced backup → final log line confirming scheduler alive
✅ v246.21b — News Reaction Tape rebuilt to match mockup — removed Source column, removed clickable filter pill row, collapsed to 8 columns (TIME · HEADLINE · ASSETS · IMPACT · 1M · 5M · 15M · STATE) · STATE column shows descriptive phrases from new _classify_persistence_phrase() helper: "Faded risk premium", "Risk-on follow-through", "Rates cap intact", "Driver relationship intact", "Haven bid holds", "Crude impulse holds", "Partial recover" · phrases are asset-keyed (NQ/ES get rates/driver language, GC gets haven language, CL gets crude language) · "Impact ≥ medium" shown as static rounded badge in panel header, always filtered to medium+ server-side · weekend-aware empty state: detects getUTCDay() === 0 || 6 and shows "Weekend lull — no high-signal headlines with measurable market reaction. Feed resumes when futures reopen." · airy row padding 16px + font 13px to match mockup typography · bottom disclaimer line restored
✅ v246.21a — News Reaction Tape timestamp parser fix — _parse_news_ts was trying to parse item['ts'] as a full datetime but the aggregator stores ts as HH:MM string and ts_sort as unix int · rewrote parser to accept the full item dict and read ts_sort first (authoritative), fall back to HH:MM + today's date, final fallback to published/time/date/published_at/pub_date as ISO · first two cache ticks returned 0 rows because driver matrix hadn't fetched asset bars yet; third tick populated 10 rows
✅ v246.21 — Command Surface Stage 6: News Reaction Tape — new _NEWS_REACTIONS_CACHE + 2-min worker · walks headlines from _NEWS_CACHE, keyword-matches to affected assets (BTC: bitcoin/btc/crypto; ES: s&p/stocks; NQ: nasdaq/tech; GC: gold/bullion; CL: oil/crude/opec/wti/brent; plus Fed/CPI cross-linking to all risk assets) · measures reaction via _measure_reaction() on cached asset bars from _DRIVER_MATRIX_CACHE.asset_bars (zero new yfinance calls) at T+0/T+5m/T+15m · best-asset selection picks the asset with largest |T+15m| magnitude and marks it · Impact = High if |T+15m| ≥ 0.5%, Medium ≥ 0.15%, Low otherwise · server-side filter via ?min_impact=medium|high · caps at 25 rows sorted most-recent first · worker starts 15s after boot (after news aggregator + driver matrix caches)
✅ v246.20 — Command Surface Stage 5: Cause & Effect — 3-tab panel (Lead-Lag Timeline / Impulse Map / Driver Overlay) · extended _DRIVER_MATRIX_CACHE to persist raw asset_bars + driver_bars as [(ts_iso, close), ...] tuples (last 60 bars each) so C&E endpoint needs zero new yfinance calls · new helpers: _detect_impulses() (rolling z-score > 1.8σ on 5m log returns), _asset_response_after() (cumulative return over N bars forward with "held" flag), _build_cause_effect(asset) (composer) · asset selector pills BTC/ES/NQ/GC/CL (default NQ) · narrative box above tabs with trader-voice interpretation built from timeline + impulse map counts · Impulse Map classifies each reaction as Held / Accelerated / Faded / Reversed / Unclear · Driver Overlay SVG chart (900×260 viewBox) with normalized asset line (solid amber) + best-correlated driver (dashed blue) since session open · endpoint /api/command-surface/cause-effect?asset=NQ recomputes on each call from cached bars
✅ v246.19 — Command Surface Stage 4: Risk Rail — 6 sticky cards sorted client-side by severity: Next Event (scans 2026 high-impact event list, countdown + affected assets, Imminent/Today/Tomorrow/Upcoming states) · Headline Risk (reads _NEWS_CACHE, buckets last 120 headlines into topic clusters Rates/Geopolitics/Earnings/Crude/Crypto/China, intensity 0-100) · Vol Shift (new _compute_realized_vol_2h helper, ES 5m bars, annualized √(252×78), compared to 90d baseline from _MARKETS_AUG_CACHE) · Correlation Break (weakest strength macro pair from Driver Matrix cells) · Invalidation (focus asset closest to VWAP from watchlist cache) · Account Heat (today's P&L from trade_results vs $700 default daily cap) · new _RISK_RAIL_CACHE + 2-min worker starting 12s after boot · six composer functions each wrapped in own try/except so one card failing can't kill others · sev-high adds red border + pink-tinted bg, sev-med adds amber border
✅ v246.18a — Driver Matrix push direction sign fix — root cause: push formula had d_move < 0 → push = exp (wrong sign) · correct formula is push = exp × sign(d_move): DXY up + BTC inverse (-1) = push -1 (pressure), DXY down + BTC inverse (-1) = push +1 (tailwind) · BTC:DXY cell was reading "DXY pressure" with direction negative when DXY was down -0.16%, which contradicted the inverse relationship · also renamed label vocabulary "supportive" → "tailwind" for traderly voice · confirmed fix via curl: BTC:DXY now reads direction positive / label "DXY tailwind" / tooltip shows "DXY -0.16% session · corr +0.19 (inverse) · lead"
✅ v246.18 — Command Surface Stage 3: Driver Matrix — live 5×7 grid (BTC/ES/NQ/GC/CL × DXY/US10Y/VIX/Breadth/Volume/News/Trend) · each cell: direction arrow (↑/↓/↔), strength N/3 (from |Pearson corr| bucketed), lead/sync/lag badge (from cross-correlation time-shift ±3 5-min lags), label text, hover tooltip with raw correlation + session % move · new helpers: _pearson(), _cross_corr_lag(), _returns_from_series(), _strength_from_corr(), _lead_lag_state() · EXPECTED prior table (BTC/ES/NQ inverse to DXY/US10Y/VIX, GC haven=+1 to VIX, CL neutral to rates/vol) · Breadth cell from NQ vs ES leadership, Volume from watchlist cache rel_volume, News from cached aggregator keyword match, Trend from range position + session pct · _DRIVER_MATRIX_CACHE + 5-min worker starting 10s after boot · only new yfinance load is 3 macro driver intraday fetches (DX-Y.NYB, ^TNX, ^VIX) · cell background tint green/red/slate by direction for visual scan
✅ v246.17e — Regime Strip aesthetic final pass — user flagged that pills had inverted styling (solid bg + black text) vs the mockup (tinted bg + bright colored text + matching border) and bars were solid orange vs mockup's top-orange-bottom-dark gradient · rewrote all 7 pill classes: .p-green/.p-amber/.p-peach/.p-salmon/.p-red/.p-orange/.p-slate now use rgba(color, .15) bg + solid color text + rgba(color, .35) border · bars now use linear-gradient(to bottom, #fbbf24 0% → #f59e0b 35% → #b45309 75% → #78350f 100%) matching the 3D lit-from-above look of the mockup · empty placeholder bars use the same gradient at 14%/10% alpha · 5th iteration to finally match the mockup exactly
✅ v246.17d — Regime Strip scaling fix — sparkline scaling rewritten to enforce a minimum 25-point spread: if observed score range is <25, the renderer expands symmetrically around the midpoint so bars always show visible variation even when scores move tightly (4th iteration) · bar color #f59e0b → #e99110 for richer saturation · pill padding 4px 10px → 6px 14px (chunkier) · pill border-radius 5px → 4px · header text 11px → 13px for "RISK APPETITE" presence · driver text 12px → 13px with 1.55 line-height for 2-line wrapping · max bar height 28px → 42px so peaks fill the sparkline zone
✅ v246.17c — Regime Strip historical backfill — new _backfill_regime_history() fetches 1-day 5m bars for ^VIX, DX-Y.NYB, ^TNX, ES=F, NQ=F, BTC-USD, GC=F, CL=F and walks backward through the last 18 historical ticks running the same scoring functions at each point-in-time · appends 18 real historical score points to each card's history deque BEFORE the worker loop starts so sparklines show real shape from first render · locks in the "open score" baseline from the first historical point so delta vs open is meaningful from first live tick · pill shape rewritten to border-radius: 5px soft rectangle (was fully round 999px) with padding: 4px 10px · card proportions: min-height 240px → 280px, padding 20px → 22px, score size 48px → 52px with -2px tracking · 8 one-time historical yfinance fetches at boot, normal 2-min cycle after
✅ v246.17b — Regime Strip design iteration 2 — card background rgba(20,24,33,.6), border rgba(148,163,184,.15), border-radius 10px, padding 20px/22px, min-height 200→240px · new _state_pill_class() function maps each card's state vocabulary to the right pill color (Risk On→green, Expanding→amber, Soft→pink, Restrictive→red, Oil-led→amber) · sparkline renderer fixed to 18 bars with ghost placeholders from first render (no more flat-bar look) · all 5 scorers upgraded to return 2-sentence trader-voice narrative instead of raw data readouts
✅ v246.17 — Command Surface Stage 2: Regime Strip — 5 composite regime cards with title, state pill, 52px score number, ± delta vs open, 18-bar amber sparkline, 2-line narrative driver text · scorers: _score_risk_appetite (VIX/ES/DXY/NQ-ES leadership weighted composite), _score_vol_state (VIX-banded Compressed/Normal/Expanding/Extreme), _score_dollar_pressure (DXY daily + MTD), _score_rates_pressure (UST10 level + bps change), _score_commodity_shock (max |oil%|/|gold%|) · new _REGIME_CACHE + 2-min _regime_worker starting 8s after boot · endpoint /api/command-surface/regime
✅ v246.16 — Command Surface Stage 1: Focus Watchlist — new top panel on MKT Technicals tab with 5 focus assets (BTC/ES/NQ/GC/CL) · columns: Asset · Last · % · Range · VWAP · PDH · PDL · ATR used · Rel Vol · new _COMMAND_SURFACE_CACHE + 5-min worker computing VWAP (typical_price × volume), PDH/PDL from daily bars, ATR14, relative volume vs 10d median, range position, signed %distances · endpoint /api/command-surface/watchlist · 8 intraday 5m + daily bar fetches per cycle · shaded gradient range bar (red→amber→green 8px tall) with 12px white-bordered dot marker
✅ v246.15 — CRITICAL server-authoritative P&L + 9-row heal — root cause: api_signal_state CLOSED branch had if close_price and ep and (pnl == 0 or pnl != pnl): recompute — so the server only recomputed P&L when client sent zero or NaN, and TRUSTED any non-zero client value · trade #288 was closed twice (15:01 first attempt, 15:07 retry) and the client cached stale -$120 pnl from the first attempt and resent it on the retry with the correct close_price=25280 · server trusted the -$120 → LOSS row with target_hit=1 AND stop_hit=1 simultaneously · decision_journal at 15:07:57 UTC confirmed server HAD computed correctly (+$58.50) but final DB row was wrong · Wider audit: cross-referenced setup_alerts with populated close_price vs trade_results stuck at result_status='OPEN', pnl_dollars=0 — found 9 stuck rows across 3 days (8-10 Apr) · Fix: new _compute_trade_pnl(asset, direction, entry_price, stop_loss, close_price) helper as single source of truth (BTC dynamic notional, futures multipliers GC=10/CL=100/MNQ=2.00/ES=5.0, returns pnl_dollars+points+result derived from pnl sign) · api_signal_state CLOSED branch rewritten to NEVER trust client pnl/points/result, only reads close_price + close_time, always recomputes via helper, logs any client/server disagreement · manual close path clears target_hit=0, stop_hit=0 explicitly so stale AGENT02 flags can't linger · deleted v243.1 reconcile and v244 sign-derivation as superseded · Heal: v246_15_heal_all_stale_opens.sql recomputed all 9 rows from stored close_price using the same formulas · daily reconciliation confirmed: 8 Apr +$387.66, 9 Apr +$939.76, 10 Apr +$411.34 — total recovered: +$1,738.76 previously missing P&L
✅ v246.14 — Pill uniformity + Trade History rename + YTD column + sticky LS header — stripped inline color styles from 5 dev-pills so .active orange-bg + black-text is uniform · ICT Signals renamed to Trade History · VIX/UST10 special handling: PCT_INSTRUMENTS = new Set(['VIX','UST10']) — these show dash in %Chg/%MTD/%YTD columns since they're themselves % values · added %YTD column between %MTD and 90D Vol in both Markets and Macro tables · new new_futures_ytd dict in augment cache · clock padding 9→13px, clock-time 20→24px · Investment Intelligence page wrapped title+pills+HMA sub-pills in a single sticky header matching Trading Intelligence pattern
✅ v246.13 — Un-indent tables + MKT Technicals rename — removed padding-left:var(--table-indent) from Markets/Macro imkt-body + imacro-body · first column widened 200→280px · range bars shrunk (min-width 220→140px, height 4→3px, dot 8→6px) · Market Conditions + Correlation panels moved from Overview to Trade Setups tab · Trade Setups pill renamed to MKT Technicals — this is where the full Command Surface gets built over Stages 1-6

Overview tab rebuilt:

market-cards2

Macro row also converted to same Bloomberg table format

Nikkei 225 (^N225)

Shanghai Composite (000001.SS)

/api/macro

change

high

low

price

pct

.imkt-*

Macro Calendar tab deleted

New Trade Setups tab

Overview final layout:

Data layer fix:

_get_market_snapshot

high

low

open

Routing:

showTradingTab

loadEconCalendar

startFJAutoRefresh

setups

.imkt-*

loadMacroRow

renderMacroTable

MACRO_ORDER

MACRO_COLOR

market-cards2

renderMarketCards

✅ v245 — AGENT02 MNQ multiplier + manual-close race guard — Bug 1 (the $0.50 orphan): AGENT02_trading.py CONTRACT['MNQ']['multiplier'] was 0.50 · web.py was patched to 2.00 in v225 but AGENT02 was never touched · every AGENT02-monitored MNQ trade since v225 was priced at a quarter of correct · fixed to 2.00 plus docstrings (module header line 9, comment line 35) and LLM prompt contract-specs line 665 · added CRITICAL comment above the CONTRACT dict warning that MNQ must stay in sync with AGENT00_web.py PNL_MULT · Bug 2 (the race): _check_fills_once SELECTs all open signals then iterates with 60s of I/O / price fetches / DB writes · if the user manually closes a trade (via sigClose) during that window, AGENT02's cached row list still has it and the subsequent target_hit/stop_hit write UPDATEs the row back to WIN/LOSS with AGENT02's idealized levels — clobbering the manual close · Fix: SELECT now also filters on sa.signal_state NOT IN ('CLOSED','SUPPRESSED') (belt); target_hit and stop_hit write blocks each re-read signal_state + result_status inside the write transaction and skip the UPDATE if already terminal — logging a warning for audit (braces) · Blast radius: 3 corrupted MNQ ORB rows (#271, #272, #278) all closed manually at profit then overwritten to LOSS with $0.50/pt pricing — explains the exact on-screen P&L values · Heal SQL: v245_heal_mnq_orb.sql recomputes the 3 rows from stored close_price at $2/pt with result derived from pnl sign · non-MNQ audit query returned only a $2 rounding-drift row (#124 GC) which is not worth touching · Lesson (again): this is the THIRD time an AGENT02 orphan has surfaced — v240.1 (killzone gate), v241.1 (close-path parity), now v245 (multiplier + race). Every web.py trading-discipline change MUST grep AGENT02 for the same value/pattern before shipping.
✅ v244 — Result-sign authority fix (trade #275 bug) — root cause: sigClose UI had two buttons (✓ Take Profit / ✕ Stop Hit) that hardcoded result='WIN' or 'LOSS' regardless of the actual close price the user then typed into the prompt · server api_signal_state trusted the client's result field (auto-determine only fired if missing/invalid), so clicking ✓ Take Profit then typing a close price below entry created a WIN row with negative pnl · trade #275 manifested this exactly · Server fix: api_signal_state CLOSED branch now ALWAYS derives result from the sign of the final computed pnl — never trusts client · logs a warning when the client's submitted result disagrees with the computed one (for audit visibility) · Client fix: sigClose recomputes actualResult from computed pnl sign before POSTing, and toasts a warn-level message when button intent differs from outcome · EOD monitor side-fix: the between-levels branch of the EOD hard-close block had a hardcoded result='LOSS' even when the pnl at market close was positive — fixed to derive from pnl sign too · ICT killzone force-close branch (Rule 3) already derived correctly, untouched · AGENT02 TP/SL monitor already derives from target_hit/stop_hit booleans, untouched · DB patch: v244_fix_trade_275.sql provided separately to recompute #275 from its stored close_price and flip result to LOSS
✅ v243 — BTC stale-alert sweep + legacy $0 P&L backfill — BTC sweep: new hourly block inside _monitor_open_signals (between Rule 3 killzone force-close and NY EOD hard close) suppresses BTC ALERTED/PENDING rows older than 24h with no trade_results row · closes the gap left by v240.2 BTC carve-out (weekday EOD excludes BTC because BTC is 24/7, so stale BTC alerts previously never got swept) · gated to once per hour via _monitor_open_signals._btc_sweep_last_hour function attribute so the 60s loop doesn't hammer the table · cutoff computed as datetime.now() - 24h in UK-local-naive to match alert_time storage convention · close_reason='BTC_STALE_SWEEP_24H' · each sweep emits an _auto_journal entry for audit · Legacy $0 P&L backfill: v241_change_b_pnl_backfill.sql executed on SIMBA (was pending since v241) · scoped to trade_results rows where pnl_dollars=0 AND result_status IN ('WIN','LOSS') and setup_alerts.close_price IS NOT NULL · uses identical formulas to v237 server-side rescue: futures pts×multiplier (GC=10, CL=100, MNQ=2, ES=5), BTC dynamic (close-entry) × 350 / ABS(entry-stop) × direction · 3-step workflow preview → update → verify executed cleanly
✅ v242 — Trade history delete button — per-row ✕ DEL button in the status/rationale column of the trade history table · confirms with a detailed destructive-action dialog listing every subsystem affected (history, P&L totals, Edge expectancy, Strategy Health, decision journal, memory aggregates) · new histDelete(id, asset, direction, pnl) JS calls DELETE /api/signals/<id> then automatically triggers /api/eod/recalculate-memory so TRADING:EOD:* aggregates rebuild immediately · reloads both history and live signals tables after delete · api_signal_delete endpoint extended to also DELETE from decision_journal where entity_type='TRADING' (was previously an orphan gap from v235 — every delete via the endpoint now cleans all three tables in a single transaction) · Edge tab / Performance / Strategy Health / today's P&L card all read trade_results live on each render, so they self-heal on next page load without explicit recompute · WTD summary and TRADING:WEEKLY:* memory aggregates still refresh on next weekly recalc (Saturday) — acceptable drift since weekly is lagged anyway
✅ v235 — v3 Phase 1 complete: Decision journal auto-logging (7 hooks: signal create/close/suppress, auto-close TP/SL, EOD close, deal status change) · Comps CRUD (4 API endpoints + pill tab + table with median EV/Rev & EV/EBITDA stats + add/edit modal + deal linking + one-pager comps card) · Confluence capture UI (12-option checkbox editor on ALERTED/OPEN signal cards + PATCH endpoint) · Deal scoring visual breakdown (8-component bar chart on one-pager)
✅ v234 — v3 Phase 1 continued: R-multiple backfill on all existing closed trades (pnl/350) · Decision Journal system (4 API endpoints + Developer pill UI with domain filter, log/edit/delete, outcome tracking, retrospective) · Probability-weighted EV on deals (bull/base/bear probability inputs, weighted MOIC/IRR/equity on one-pager + modal) · Deal scoring model (auto-computed 0-100 composite from 8 weighted components: team, financials, exit model, risks, market, investors, MOIC, timing) · Score column on pipeline table + KPI strip · Probability badges on scenario cards
✅ v233 — v3 Phase 1 schema foundation: 17 new columns on setup_alerts (confidence, confluence, regime tags, trade_mode, day_of_week) · 9 new columns on trade_results (R-multiple, MFE, MAE, grading) · 5 new tables (trade_overrides, strategy_changelog, decision_journal, comps, deal_reviews) · _enrich_trade_result() auto-computes R-multiple on every close · signal creation captures full context · dead RSS feeds replaced · regex-in-template-literal bug fixed

Pending ☐

☐ Back up .env to password manager / iCloud
☐ Static DHCP reservation for Mac Mini
☐ Signal backtester — 6-12 months OHLCV
☐ Signal quality filter threshold — set permanent min score after 4-6 weeks data
☐ H100 live price via Nordnet (Playwright auth)
☐ AGENT05 dedicated agent file · macro price alerts
☐ Multi-turn Ask Anything — rolling conversation context
☐ Weekly trading summary auto-written to memory
☐ H100 Advisory invoicing module
☐ Contacts — link to todos/agenda/email
☐ PFIN — offsite FX rate fallback if yfinance unavailable
☐ Investment Intelligence — iPhone responsive layout
☐ Deal import: store deck text to memory for Ask Anything
☐ PNG export — add TAM/SAM/SOM pills + pedigree
☐ Weekly trading summary per-strategy breakdown
☐ Investigate duplicate BTC signals #145/#146 (same entry/close/P&L)
☐ MCL — no risk cap enforced on crude oil (ATR-based but uncapped)
☐ Auto-close monitor: write close_price to setup_alerts on auto-close
~~Decision journal auto-logging~~ ✅ shipped v235
~~Comps UI~~ ✅ shipped v235
~~Deal scoring visual breakdown~~ ✅ shipped v235
~~Confluence/confidence capture UI~~ ✅ shipped v235

Dev To-Do

Loading…

Version History

v257 current

Loading…

Loading database stats…

💾 Backup Status

Loading backup stats…

📁 Available Backups

Loading…

📓 Decision Journal

Loading journal…

System

Process · logs · quick actions

Status

Quick Actions

Recent Events (24h)

agent00.log (last 100 lines)

Personal Finance

Net worth · assets · accounts · investments

Company Financials

Hugesky

Financial Year

FY 2023

Total Assets

Total Liabilities

Net Income

Equity

Balance Sheet — Assets

Cash & Equivalents

Loans & Receivables

Investments & Financial Assets

Tangible Fixed Assets

Intangible Assets

Other Assets

Total Assets

$0.00

Liabilities

Accounts Payable

Long-Term Debts

Other Liabilities

Total Liabilities

$0.00

Shareholders’ Equity

Share Capital

Retained Earnings

Total Equity

$0.00

Income Statement

Revenue

Cost of Sales

Gross Profit

$0.00

Expenses

Operating Expenses

Other Expenses

Income Tax Expense

Total Expenses

$0.00

Net Income

$0.00

Gross Profit − Total Expenses

All figures in USD. Auto-saves to database on input change.

🎲 SCENARIO ANALYSIS

Horizon: yrs | Show in today's money

⚠ Balance-sheet only — income flows, tax (CGT/income), and mortgage paydown are not modeled. Property leverage is correctly applied (property × rate, equity = property − mortgage held constant). Edit any growth rate to re-compute live.

📈 WEALTH HISTORY (DE)

✦ AI Financial Analysis

Click ↻ Generate Analysis to run AI financial analysis on demand.

ASSETS
Cash	$461,681
Accounts Receivable	$389,438
Other Current Assets	$54,215
Fixed Assets	$226,857
Other Assets (leases, deposits)	$707,962
Total Assets	$1,840,153

LIABILITIES & EQUITY
Accounts Payable	$1,045,672
Credit Cards	$96,788
Operating Lease Liabilities	$520,822
Crowdfunding + Royalties + Other	$250,719
Total Liabilities	$1,914,001
Paid-In Capital + Common	$21,562K
Retained Earnings	($20,163K)
Total Equity	($73,848)

	2023			2024			2025			FORECAST ONLY
	Forecast	Actual	Var	Forecast	Actual	Var	Forecast	Actual	Var	2026F	2027F
Revenue ($M)	14.9	15.5	+4%	19.7	20.2	+3%	25.5	26.0	+2%	30.3	31.2
EBITDA ($M)	—	3.6	—	—	5.4	—	—	6.5	—	—	—
EBITDA Margin	—	23%		—	27%		—	25%		—	—
Net Income ($M)	3.3	—		3.7	—		7.6	—		11.7	11.8
Net Margin	22%	—		19%	—		30%	—		39%	38%