Quant Trading Infrastructure.
Without the Infrastructure Overhead.

# strategy.py
from sklearn.ensemble import RandomForestClassifier

def train(data, config):
    features = build_features(data)
    model = RandomForestClassifier(
        n_estimators=250,
        max_depth=8,
        random_state=42,
    )
    model.fit(features.X, features.y)
    metrics = {
        "sharpe": 1.82,
        "profit_factor": 1.41,
        "max_drawdown": 0.084,
    }
    return model, metrics

def predict(model, market_data, config):
    point = latest_feature_vector(market_data)
    prob = model.predict_proba([point])[0][1]
    if prob > 0.64:
        return {"signal": "UP", "confidence": prob}
    if prob < 0.36:
        return {"signal": "DOWN", "confidence": 1 - prob}
    return {"signal": "HOLD", "confidence": 0.18}