Rule Lab
Find the rule settings that actually fit how you trade. Each setting is replayed through your Ghost. The best one is picked on your earlier trades and then tested on later trades it never saw, so a lucky number cannot fool you.
1 trade looked best on the earlier 135 trades but did worse than your current 5 trades on the later 59. That is what curve-fitting looks like. Keep your current setting.
Ghost net P&L at each setting, split into the trades used to choose (blue) and the later trades held back (white). Bold labels: your current setting, and in green the best on earlier trades.
- Your current setting
- 5 trades
- Best on earlier trades
- 1 trade
- Later trades, current
- +$8,851
- Later trades, best
- +$4,848
- Difference on unseen trades
- −$4,003
- You, actual (later trades)How you really traded in that period
- +$11,893
No change recommended. Adopting a setting that only won on the past is how traders overfit.
Every setting
Ghost replay, all trades| Max trades per session | Earlier | Later | All trades | Taken / skipped |
|---|---|---|---|---|
| 1 tradebest earlier | +$4,294 | +$4,848 | +$9,142 | 63 / 131 |
| 2 trades | −$580 | +$7,568 | +$6,988 | 109 / 85 |
| 3 trades | −$1,443 | +$8,689 | +$7,246 | 124 / 70 |
| 4 trades | −$1,443 | +$8,851 | +$7,408 | 125 / 69 |
| 5 tradescurrent | −$1,443 | +$8,851 | +$7,408 | 125 / 69 |
| 6 trades | −$1,443 | +$8,851 | +$7,408 | 125 / 69 |
| 8 trades | −$1,443 | +$8,851 | +$7,408 | 125 / 69 |
| 10 trades | −$1,443 | +$8,851 | +$7,408 | 125 / 69 |
Only this rule changes; every other rule stays as you set it. You actually made −$3,087 over all trades. Past results on a replay are not a promise; treat any change as something to test forward on small size.