Methodology: 3 folds — 6-month in-sample + 3-month out-of-sample, rolling forward by 2 months. Folds: IS May–Oct 2025 → OOS Nov–Jan · IS Jul–Dec 2025 → OOS Jan–Mar · IS Sep 2025–Feb 2026 → OOS Mar–May Param selection: Best in-sample Sharpe → its OOS metrics reported (no OOS peeking). Runtime: ~5 hours (3 × full grid run) — designed to run overnight. ⚠ Do not run Grid Backtest simultaneously.
Checking…
CPU Workers per fold
same setting as Grid Backtest
Walk-Forward Aggregate — Best IS-Chosen OOS Metrics
Each cell = OOS performance of the param set that ranked #1 in-sample for that fold.
Mean/Std across 5 folds. No OOS snooping.
Cap
Vol
Folds ✓
Mean OOS Sharpe
Std
Min → Max
Wt Win Rate
Wt Avg Ret
Total N
Per-fold Sharpes
Per-Fold Detail
Fold
Cap
Vol
IS Sharpe
OOS Sharpe
OOS Win%
OOS AvgRet
OOS N
Chosen Params
Signal: Bullish engulfing → long · Bearish engulfing → short Grid: 1,296 entry combos × 120 exit combos × 6 segments · IS before Jul 2025 / OOS from Jul 2025 Filters: entry delay · body ratio · volume · trend (MA50/200) · RSI · dist from MA Target: 5%–100% · Runtime: ~20–30 min