5 năm · Full cost model (spread, slippage, commission, swap) · Sensitivity sweep · Walk-forward
| # | Variant | Trades | WR | Avg R | Total R | Return % | Max DD | PF | Final $ |
|---|---|---|---|---|---|---|---|---|---|
| Tháng | Trades | WR% | R | % P&L | $ P&L | Equity |
|---|---|---|---|---|---|---|
Cùng params (risk=1.5%, RR=2.0). Nếu test perform gần bằng train → strategy robust, không overfit.
| Symbol | Train (2021-04 → 2025-01) | Test OOS (2025-01 → 2026-04) | Degradation | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Trades | WR | Return | Max DD | Trades | WR | Return | Max DD | ||
Annualized test return (1.25 năm):
Mỗi symbol → grid 3×4 (risk × RR). Màu = Return %.
| Symbol | Test 1.25y | Annualized | Max DD |
|---|
symbols: [EURUSD, GBPUSD, XAUUSD, NAS100, BTCUSD] entry_tf: M15 htf_trend: D (EMA50 & EMA200) structure_tf: H4 (BOS, lookback=3 swings) fvg_window: 48h TTL session_filter: UTC 7-22 risk_per_trade: 1.5% max_positions: 2 (across all symbols combined) max_daily_dd: -10% take_profit: RR 2.5 base, RR 3.0 cho BTCUSD/EURUSD (khi data ủng hộ) stop_loss: FVG low/high ± 0.3 × ATR(14) partial_tp: OFF (backtest cho thấy giảm lợi nhuận) trailing_stop: OFF cost_model: spread + slippage + commission cố định per instrument kill_switch: -15% equity trong tháng → pause 1 tháng