트레이딩 봇 만들어보고 싶은데 같이 해볼 분 있나요? 🤖

I’m interested in collaborating on a regime-adaptive trading system rather than a single fixed strategy.

The core idea is to decompose market behavior into latent regimes using a Hidden Markov Model trained on volatility, return distributions, and volume imbalance. Each regime activates its own execution logic instead of relying on one universal entry rule.

Signals come from the confluence of three independent components: a microstructure momentum module based on intrabar volatility expansion and order-flow proxies, a higher-timeframe directional bias estimated via a Kalman filter trend model, and a mean-reversion exhaustion detector using entropy compression and volatility-adjusted z-scores.

Trades are only taken when all components align within the same inferred regime.

Risk management is dynamic. Position sizing is volatility-scaled and capped using drawdown estimates from Monte Carlo simulations, with volatility-based stops that adapt per regime. Trade management favors partial exits driven by changes in regime probability rather than fixed price targets.

The system uses walk-forward validation with parameter stability constraints and is evaluated on risk-adjusted metrics like Sortino ratio, MAE, and regime-conditioned expectancy. The goal is a transparent, modular architecture that can be stress-tested and evolved, not a black-box optimizer.