Strategy — MM Positions Analyzer

Data Check

Fetch 1m data if needed, check plan fillability

➜

Sim Execution

Execute fills, close positions, score outcomes

Account

➜

Feedback Learning

Learn from outcomes, adjust confidence

➜

Analyze

Brain signals with updated knowledge

➜

Plan Next Day

Score recs, select plans for tomorrow

➜

Summary Today

Generate daily snapshot + alerts

↺

Chain: -- Last run: -- OHLCV: -- Opt OHLCV: -- Gaps: -- Replay from

▶

Pending Execution Plans

Instrument	Dir	Expiry	Orders	Debit / Credit	Stop	Target	Score	Conf	Win Rate	Accts	Status
Loading daily plan...

Risk Profile

Click a plan row to see P&L curve

Yesterday's Scorecard

Paper Account View Full Account

Date	Ticker	Expiry	Strategy	Dir	Conf	Price	Regime	Status	1D	2D	5D	Verdict
Click Load to fetch recommendations.

Strategy Performance by Regime

Strategy	Regime	Direction	Total	Correct	Wrong	Partial	Win Rate	Avg 1D %	Avg 5D %	Conf Adj
Loading performance data...

Confidence Adjustments (Learned)

Strategy	Regime	Adjustment	Reason	Sample Size	Updated
No adjustments yet. The feedback loop runs weekly.

Macro Market State Cross-ticker market regime from engine/macro_engine.py

Loading macro state...

Scoring Weights Controls how the Brain ranks recommendations for execution

Each recommendation's composite score = base confidence × weight + DTE bonus × weight + regime conviction × weight + ... Higher weight = more influence. Set to 0 to disable a factor entirely.

base

Strategy confidence (1-3 stars)

dte

DTE sweet spot bonus (7-30d optimal)

regime

Regime conviction (direction × GEX/DEX alignment)

win_rate

Historical win rate by (strategy, regime)

learned

Feedback loop confidence adjustments

corr_penalty

Penalty for holding same ticker already

cost

Option premium cost vs underlying price

macro

Macro market regime alignment

min_score

Min score to generate an execution plan

max_plans

Max execution plans per day

max_same_dir

Max same direction in one batch

max_same_strat

Max same strategy type in one batch

Shadow Candidates All strategies evaluated for each ticker — winners + runner-ups with outcome scores

Ticker: Date:

Select a ticker and click Load to see all evaluated strategies.

Strategy Catalog From web/data/strategies.json — conditions the Brain evaluates at each decision point

Loading strategies...

Regime Classification 7 GEX regime cases from regime_classification.json

Loading regimes...

Judgement Criteria & Human Knowledge Dealer mechanics, signal rules, case studies, industry wisdom

Loading criteria...

Trade Structure Models Vol/OI flow detection patterns from trade_structures.json

Loading trade structures...

Signal Validation Rules How prediction outcomes are scored (validation_rules.json)

Loading validation rules...

GEX × DEX Trend Matrix Directional implications from GEX sign + DEX sign combinations

Loading matrix...

Evaluation State Fields All ~85 features the Brain computes at each decision point (evaluator.py build_state) — stored as ML training data

Field	Type	Source	Description
REGIME FLAGS (original)
netPos	bool	GEX	Net GEX ≥ 0 (positive gamma regime)
aboveFlip	bool	GEX	Price above GEX flip level
belowFlip	bool	GEX	Price below GEX flip level
noFlip	bool	GEX	No flip level found (persistent regime)
pinnedAtMaxG	bool	GEX	Price within 0.5% of max gamma strike
nearFlip	bool	GEX	Price within 1.5% of flip level
dexBullish / dexBearish	bool	DEX	Net DEX > 0 / < 0
ivHigh / ivElevated / ivLow	bool	IV	ATM IV > 40% / > 25% / < 15%
dte	int	Calendar	Days to expiration
resistAbove / trapBelow / supportBelow	float	GEX levels	Nearest +GEX/−GEX levels
oiSkewCall / oiSkewPut	bool	OI	P/C OI < 0.67 / > 1.5
gexConcentrated	bool	GEX	Max \|GEX\| > 4× average
ENRICHED REGIME (7-case + matrix)
regimeCase	int 1-7	Classifier	Regime case (1=Pinned, 2=Near Flip, 3=+GEX Above, 4=+GEX Below, 5=−GEX Above, 6=−GEX Below, 7=No Flip)
regime1_pinned … regime7_noFlip	bool	Classifier	One-hot regime flags for ML
dealerMode	+1/−1	GEX	+1 stabilizing (−GEX), −1 amplifying (+GEX)
matrixBullish / matrixBearish	bool	GEX×DEX	From gex_dex_matrix.json implication
matrixOscillating	bool	GEX×DEX	+GEX/−DEX = oscillating range
PRICE POSITION (relative features)
posGex1 / posGex2 / negGex1 / negGex2	float	GEX levels	±GEX #1 and #2 strike levels
rangePosition	0-1	Computed	Price position within [−GEX#1, +GEX#1] range (0=low, 1=high)
rangeWidthPct	%	Computed	Width of GEX range as % of price
nearRangeHigh / nearRangeLow	bool	Computed	Price in top/bottom 20% of GEX range
inRangeMiddle	bool	Computed	Price in middle 30% of GEX range
distToFlipPct / distToMaxGPct	%	Computed	\|price − flip\| / price, \|price − maxG\| / price
flipInRangePct / maxgInRangePct	0-1	Computed	Flip and maxG position within GEX range
VEX & CONVERGENCE
vexBullish / vexBearish	bool	VEX	Net VEX > 0 / < 0
allThreeBullish / allThreeBearish	bool	GEX+DEX+VEX	All three exposures aligned bullish/bearish
GEX PROFILE SHAPE
gexPeakRatio	float	GEX curve	Max \|GEX\| / avg \|GEX\| — peak concentration
gexVeryConcentrated / gexModeratePeak / gexSpreadOut	bool	GEX curve	Peak ratio > 4 / 2-4 / ≤ 2
gexBimodal	bool	GEX curve	Second peak > 60% of primary peak
gexPeakAbovePrice / gexPeakBelowPrice / gexPeakAtPrice	bool	GEX curve	Peak strike relative to current price
gexHHI	0-1	GEX curve	Herfindahl concentration index
gexSkewness	float	GEX curve	3rd moment — distribution asymmetry
gexAsymmetryRatio	float	GEX curve	\|neg GEX sum\| / \|pos GEX sum\| — mean-reversion predictor
asymmetryBullish / asymmetryBearish	bool	GEX curve	Asymmetry ratio > 1.3 / < 0.7
GAMMA TILT & IV SKEW
gammaTiltPct	0-1	Gamma	Fraction of gamma mass above price (0.5 = symmetric)
tiltCallHeavy / tiltPutHeavy / tiltSymmetric	bool	Gamma	Tilt > 0.65 / < 0.35 / 0.35-0.65
ivSkewRatio	float	IV	ATM call gamma / put gamma — skew proxy
ivSkewNormal / ivSkewInverted / ivSkewBalanced	bool	IV	Ratio > 1.15 / < 0.87 / between
P/C RATIO LEVELS
pcrExtremePut / pcrBearish / pcrNeutral / pcrBullish / pcrExtremeCall	bool	OI	P/C OI classification (5 buckets)
GAMMA SQUEEZE & TRADE STRUCTURES
gammaSqueeze	bool	Squeeze engine	DEX+GEX+VEX convergence detected
squeezeBullish / squeezeBearish	bool	Squeeze engine	Squeeze direction
posGammaRegion	bool	GEX engine	Price between flip and +GEX#1 — dealer stabilizing zone
posGammaRegionWidthPct	float	GEX engine	Width of +γ zone as % of price (narrow=spring, wide=magnet)
posGammaRegionPosition	float	GEX engine	Position in zone: 0=at flip, 1=at +GEX#1
gammaSupportReclaim	bool	Composite	SP-1: price in +γ zone near floor + 2+ confirms (squeeze, greeks, concentrated GEX)
hasTradeStructure	bool	Structure engine	Institutional flow patterns detected
tradeStructureBullish / tradeStructureBearish	bool	Structure engine	Structure direction
tradeStructureSpread	bool	Structure engine	Spread pattern (not just directional)
SHAPE VECTORS (Neural Net)
_gexShape / _dexShape / _vexShape	float[20]	Interpolation	20-point normalized curve ±10% of price — for CNN/neural net training
MACRO STATE (cross-ticker context)
macroRiskOn / macroRiskOff / macroCrisis / macroTransition	bool	Macro engine	Composite regime classification from VIX + breadth + momentum + haven flow
macroVixElevated / macroVixExtreme	bool	VIX	VIX > 20 / > 30
macroVixPeakReversal	bool	VIX	VIX dropped > 15% from 20d peak — buy-the-dip signal
macroBroadAdvance / macroBroadDecline	bool	Breadth	≥ 70% / ≤ 30% of sectors above SMA20
macroMomentumBull / macroMomentumBear	bool	SPY	SPY 5d/20d returns indicate bullish/bearish momentum
macroTickerSectorLeading / macroTickerSectorLagging	bool	Sector RS	This ticker's sector ETF in top 3 / bottom 3 by 5d return vs SPY
_macroRegimeScore	float −1..1	Composite	Weighted composite: −1 = crisis, +1 = strong risk-on
_macroVixLevel / _macroVix5dChange / _macroVixFromPeakPct	float	VIX	Current level, 5d change %, % from 20d peak
_macroBreadthPct / _macroSpy5dRet / _macroSpy20dRet	float	Market	Breadth %, SPY returns, momentum
_macroHavenMomentum / _macroVixGldCorr / _macroSectorDispersion	float	Cross-asset	Haven flow, VIX/GLD correlation, sector return dispersion

Trader Profiles (Tournament) Predefined weight profiles for paper account competition

Loading profiles...

Feature Pipeline & Training Data intelligence.db — feature extraction for future ML training

--Snapshots

--Outcomes

--Human Labels

--Feature Fields

The feature pipeline captures the full market state at each decision point (regime, GEX/DEX/VEX levels, IV, OI, signal flags) along with actual price outcomes. This is the training data for future ML models (XGBoost, neural net, RL agent). More data = better models. Every pipeline run adds to this dataset.

Feedback Control Loops Multi-loop cybernetic architecture (Tsien model)

Loop 1: Execution

minutes–hours

Monitor positions → stop/target/expiry check → close/adjust → update portfolio

trader.py → review_open_positions()

Loop 2: Strategy Adaptation

days–weeks

Track per-strategy win rate → adjust signal confidence → detect regime shifts → update knowledge

feedback.py → calibrate_confidence()

Loop 3: Model Evolution

weeks–months

Retrain ML models → evaluate vs baseline → promote better model → add new features → backfill → retrain

Planned — requires intelligence.db feature store