Drawdowns deeper than 8%, 2021–2026
Every peak-to-trough decline deeper than 8% in the backtest, with its depth, duration and recovery.
- Maximum drawdown
- −18.5%
- Longest under water
- 241 days
- Losing months
- 40%
Findings
- 12 drawdowns deeper than 8% in 5.7 years, about two a year.
- The longest period without a new equity high was 241 days, from 13 March to 10 November 2024.
- 27 of 68 months (40%) were negative, and 3.2% of 4-hour closes set a new equity high.
Method
CAGR and maximum drawdown each reduce the path of returns to a single figure. To show the sequence of losses an allocator would have held through, every decline of more than 8% from a running equity peak was extracted from the backtest equity curve, with its depth, the days from peak to trough and the days from trough back to the prior peak.
Results
| Peak | Trough | Depth | Days to trough | Days to recover |
|---|---|---|---|---|
| 13 Mar 2024 | 1 Jul 2024 | −18.5% | 109 | 132 |
| 18 Jun 2022 | 9 Oct 2022 | −17.9% | 112 | 68 |
| 6 Jun 2026 | 19 Aug 2026 | −15.3% | 73 | 2 |
| 5 Feb 2026 | 27 May 2026 | −14.8% | 110 | 9 |
| 24 Jan 2023 | 4 Jun 2023 | −13.0% | 131 | 39 |
| 22 Aug 2023 | 19 Oct 2023 | −12.4% | 57 | 4 |
| 7 Apr 2025 | 4 Jul 2025 | −12.2% | 88 | 13 |
| 18 Sep 2025 | 13 Oct 2025 | −12.0% | 25 | 110 |
| 24 Feb 2022 | 22 Apr 2022 | −10.6% | 56 | 17 |
| 8 Dec 2024 | 14 Jan 2025 | −9.7% | 36 | 43 |
| 28 Feb 2025 | 2 Apr 2025 | −9.5% | 33 | 4 |
| 23 May 2021 | 17 Jun 2021 | −8.1% | 24 | 5 |
- Frequency. Twelve drawdowns exceeded 8% in 5.7 years. One or two in a given year would be consistent with the backtest.
- Duration. The median decline took 65 days to reach its trough and the longest 131. Recoveries took between 2 and 132 days, typically arriving in bursts when a new trend ran.
- Time under water. After the 13 March 2024 peak, equity took 241 days to set a new high, on 10 November 2024. New highs were set on 3.2% of 4-hour closes, and equity stood more than 10% below its running peak about a fifth of the time.
- Monthly. 27 of 68 months lost money (40%). Worst month −8.8% (July 2026), best +30.5% (November 2024). Longest losing streak three months. Six of 22 full quarters were negative.
Implications
This is the usual return profile of trend following, which is positively skewed: frequent small losses on trends that fail, offset by a small number of large gains on trends that persist. Exiting losing trades sooner also cuts off some trends before they develop, and take-profit orders that cap the winners tested worse in the backtest.
Four of the twelve declines took between 109 and 131 days to reach troughs of −13.0% to −18.5%. A copy closed near one of those troughs would have realised the loss and missed a recovery that took between 9 and 132 days. An allocation is suitable only if it is sized so that a decline of that depth and length can be held.
Limitations
These are backtest drawdowns, measured on one sample with the benefit of hindsight; a future drawdown can be deeper or last longer. Including funding deepens the maximum drawdown from −18.5% to −18.8% (Note 01).