Maximise risk-adjusted return (Sharpe/Calmar) on NZD/USD 15-min data. XGBoost chosen for its strong performance on tabular financial data. Moderate depth (4) and high regularisation (gamma, alpha, lambda) prevent overfitting on a relatively small forex dataset. Subsample + colsample_bytree add stochastic diversity. SL 0.5% / TP 1.0% gives a 1:2 R:R ratio to support positive expectancy even with a sub-60% win rate. Threshold 0.55 filters marginal signals while keeping trade fr

Maximize risk-adjusted return (Sharpe/Calmar) on EUR/USD 15-min data. SMA triple-stack (20/50/200) provides trend context; supplementary momentum (RSI, MACD, Stochastic), volatility (ATR, BB), and mean-reversion (CCI, Williams %R) features give the XGBoost model a rich multi-regime signal set. XGBoost chosen for its ability to rank feature importance and handle non-linear interactions. Shallow trees (max_depth=4) with high n_estimators and low learning_rate reduce overfitting

Maximize risk-adjusted return (Sharpe/Calmar). GradientBoosting chosen for its strong performance on tabular data with structured non-linear interactions. Shallow trees (depth=4) with high n_estimators and low learning_rate reduce overfitting. subsample=0.75 adds stochasticity. Early stopping via n_iter_no_change avoids over-training on the 15-min EURUSD regime. SL=0.5%/TP=1.0% gives 2:1 reward-risk, consistent with a squeeze-breakout edge. session_filter [6,20] UTC captures