RL Strategy Discovery

Run RL training

Full 5m history, auto walk-forward folds. State: technicals + causal SMC. Training reward: LSTM next-bar return learning (blended with market). OOS scored on market PnL vs buy-hold & MA.

Timesteps / fold Folds Auto-tune folds

Hybrid: FVG+structure × logistic

Policy	OOS ret	Sharpe	Max DD	Trades	Score

Walk-forward results

Fold	Train bars	Test bars	PPO return	Sharpe	Max DD	Trades	vs B&H

Analysis charts

RL Strategy Discovery

Run RL training

Hybrid: FVG+structure × logistic

Walk-forward results

Analysis charts

OOS return — PPO vs baselines (per fold)

PPO OOS return by fold

PPO OOS Sharpe by fold

Max drawdown by fold

Win rate by fold

Trades per fold

PPO equity curves (OOS test segments)

Full-sample baseline comparison

Summary