These checks ask whether the validation result survives more expensive execution and whether performance depends on one unusually favourable part of the history.
Execution-cost stress test
| Validation metric | Modeled costs | Stress replay |
|---|
| Total return | 51.12%USD +10,359.00 | 51.12%USD +10,359.00 |
|---|
| Annualized return | 13.18%/yr | 13.18%/yr |
|---|
| Maximum drawdown | -5.55%USD -1,683.01 | -5.55%USD -1,683.01 |
|---|
| Profit factor | 2.34 | 2.34 |
|---|
| Trades | 45 | 45 |
|---|
Consistency across time
| Slice | Dates | Trades | Net result | PF |
|---|
| Period 1 | 2012-08-05 to 2016-02-07 | 8 | USD 341.56 (3.42%) | 1.56 |
|---|
| Period 2 | 2016-02-07 to 2019-08-10 | 44 | USD 5,597.92 (54.13%) | 2.58 |
|---|
| Period 3 | 2019-08-10 to 2023-02-11 | 43 | USD 4,718.55 (29.60%) | 1.83 |
|---|
| Period 4 | 2023-02-11 to 2026-08-14 | 47 | USD 9,966.41 (48.24%) | 2.23 |
|---|
What held up
- The later validation period returned +51.12% across 45 closed trades.
- Recorded spread was doubled. Commission remained USD 0.00. No validation trade outcomes changed, so the return was unchanged at +51.12%.
- 4 of 4 active chronological periods were profitable.
Watch-outs
- The validation sample is still moderate at 45 closed trades.