Autonomous Trading Intelligence

Trade Smarter.
Not Harder.

Multi-engine autonomous trading system combining ML predictions, ICT methodology, and real-time market microstructure analysis.

Access Dashboard How It Works
0
Trading Engines
0
State Machines
0
ML Features
0
Active Symbols
24/7
Autonomous
Capabilities

Multi-Layer Intelligence

ML
ML Ensemble Scoring
9 technical analyzers compute 520+ features per timeframe. Heuristic + ensemble models score every signal through 28 quality gates.
ICT
ICT State Machines
14 reversal and continuation state machines detect liquidity sweeps, fair value gaps, and structure shifts across multiple timeframe combos.
WS
Real-Time Data Pipeline
WebSocket tick data feeds candle builders, momentum scanners, and position monitors. Sub-second SL/TP execution on every price tick.
AI
AI Copilot
Event-driven session analyst powered by Claude. Extracts session bias from trade decisions and provides real-time market intelligence packets.
VC
Adaptive Volatility
VolContext framework dynamically scales thresholds across all analyzers based on current regime. Fast-path detection catches crashes in 1-2 candles.
SM
Shadow A/B Testing
6 shadow machines run modified parameters in parallel, comparing outcomes against production without risking capital. Data-driven optimization.
Architecture

Signal Flow

// Market Data Pipeline
WS Ticks → CandleBuilder + MomentumScanner
  ├── CandleClose → scalp_decision() | swing_decision()
  ├── Momentum  → adaptive threshold (0.25 × ATR_pct_5m)
  └── Tick      → position_monitor (sub-second SL/TP)

// ML Scoring (in-process, ~400ms)
9 Analyzers → 520+ features → EliteEnsemble → 28 gates

// Execution
Signal → TradeGate → RiskManager → OrderExecutor → Exchange
Trading Engines

Four Engines. One System.

MEAN
Python • ML Heuristic
  • 4 strategies (Alpha, Beta, C, D)
  • Scalp + Swing + Trend Cascade
  • 520+ feature ensemble scoring
  • Dynamic 15-symbol universe
ICT
Python • State Machines
  • 14 trading + 6 shadow machines
  • Reversal + Continuation arms
  • Rich conviction scoring
  • Draw On Liquidity system
ULIM
Rust • Tick-Driven
  • 3 combos × 2 symbols
  • Sub-second tick SL/TP
  • 5.4MB RSS memory footprint
  • Checkpoint/restore with gzip
COPILOT
Python • Claude AI
  • Event-driven pub/sub
  • Session bias extraction
  • Intel packets per symbol
  • ~$0.62/day API cost