
About
An algorithmic trading bot built for multi-symbol execution with a live production strategy and a shadow AI chart brain for evaluation. It combines rule-based trade management, broker integration, feedback-driven learning, and observability tooling to improve decision quality without risking the live engine prematurely.
Tech Stack
PythonMetaTrader 5FastAPILightGBMReactNext.jsTypeScriptPowerShell
Info
- Category
- Bot
- Status
- LIVE
- Technologies
- 8
Architecture
Code Preview
Pythonshadow_brain.py
1class ShadowBrain:2"""Shadow AI evaluator running parallel to live engine."""34def __init__(self, model_path: str):5self.model = lgb.Booster(model_file=model_path)6self.live_engine = LiveEngine()78async def evaluate_signal(self, symbol: str):9features = self.extract_features(symbol)10ai_score = self.model.predict([features])[0]11live_decision = self.live_engine.get_decision(symbol)1213# Log disagreements for learning14if ai_score > 0.7 and not live_decision.should_enter:15await self.log_disagreement(16symbol=symbol,17ai_score=ai_score,18live_action="skip",19reason="shadow_bullish_live_skip",20)2122return ShadowResult(23score=ai_score,24confidence=self.calculate_confidence(features),25)