Blog · 8 November 2026
We restated two backtests
We found three errors in how the backtests of Old Pal 1.1 and Americano 1.1 were measured. The strategies are unchanged; only how their positions were valued and sized changed. Both pages carry the corrected figures. Black Manhattan 1.1 is unaffected.
| 2022 | 2023 | 2024 | 7-year cycle | Max drawdown | |
|---|---|---|---|---|---|
| Old Pal 1.1, published | −12.0% | +8.5% | +51.3% | +21.4% | −31.0% |
| Old Pal 1.1, corrected | −10.6% | +26.3% | +47.4% | +28.4% | −30.6% |
| Americano 1.1, published | −7.7% | +10.8% | +17.6% | +10.7% | −11.5% |
| Americano 1.1, corrected | −6.6% | +11.1% | +16.3% | +10.5% | −11.3% |
- 2023: after the 3 July early close, held positions were priced at zero, a one-day loss of about 17% for Old Pal that never happened.
- 2022: the volatility brake started cold, so the first weeks ran at full exposure into a falling market. It now warms up on the weeks before, as it does live.
- 2024: the backtest stopped a day early. 31 December, a −3.8% day for Old Pal, is now in.
Old Pal 1.1 no longer reads as a one-year story. But a correction in our favour deserves more scrutiny, not less, which is why the before and after sit side by side. The page curves now come from the same backtest we evaluate the strategies with.
Every figure here is an in-sample 2022–2024 backtest, not a live result. Nothing here is investment advice. Full disclaimer.