Market event replays: who was already out, who got caught
For 35 major market events since 2010 we replayed our rule-based strategies on the affected market: which ones were already out of the market, which ones held on, and which made money. Prices and strategy curves run through 2026-11-10; results are before trading costs.
Which strategies lost less than the market, again and again
Across 29 sell-offs (every curated event where the market fell, each counted once on its primary market) the median market move was -18.8%. In the median event 98% of strategies lost less than the market, but only 5% finished with a gain. 37 of 62 strategies lost less than the market in every single event. Regime & other strategies protected best (median return -2.3%); Momentum strategies protected least (-9.4%).
Strategy families, ranked by median return in the sell-offs
| Family | Events | Median share that lost less | Median share that made money | Median share already out | Median return in window | Median return minus market |
|---|---|---|---|---|---|---|
| Regime & other | 29 | 12% | 50% | -2.3% | +14.8% | |
| Trend | 29 | 4% | 33% | -3.9% | +14.5% | |
| Volatility | 29 | 0% | 33% | -4.6% | +13.7% | |
| Mean reversion | 29 | 7% | 93% | -4.6% | +13.7% | |
| Momentum | 29 | 0% | 25% | -9.4% | +9.3% |
The 10 strategies that most often lost less than the market
| Strategy | Family | Lost less (events) | Made money | Already out | Held on and lost less | Median return minus market |
|---|---|---|---|---|---|---|
| TREND-14 Linear Regression Channel | Trend | 29 / 29 | 23 | 10 | 19 | +24.5% |
| TAIL-01 Crisis Alpha | Regime & other | 29 / 29 | 15 | 22 | 7 | +24.1% |
| MR-01 RSI Mean Reversion | Mean reversion | 29 / 29 | 8 | 27 | 2 | +18.2% |
| MR-16 Connors RSI | Mean reversion | 29 / 29 | 7 | 25 | 4 | +17.3% |
| MR-06 Z-Score Mean Reversion | Mean reversion | 29 / 29 | 6 | 28 | 1 | +17.7% |
| MR-15 Ultimate Oscillator | Mean reversion | 29 / 29 | 3 | 25 | 4 | +10.4% |
| MR-02 Bollinger Band Reversion | Mean reversion | 29 / 29 | 3 | 26 | 3 | +13.5% |
| MR-04 Ornstein-Uhlenbeck Mean Reversion | Mean reversion | 29 / 29 | 3 | 28 | 1 | +14.1% |
| MR-10 Stochastic Oscillator Reversion | Mean reversion | 29 / 29 | 2 | 24 | 5 | +8.8% |
| COMBO-04 Trend + Mean Reversion Hybrid | Trend | 29 / 29 | 2 | 28 | 1 | +16.3% |
Counted over events where the strategy had a curve on the primary market (at least 3 events). Ranked by the share of events with a smaller loss than the market, then by number of events and by how often it made money. "Already out" = inferred from the equity curve, see method. A strategy that is rarely in the market scores well here by construction.
Read this carefully: our strategies can step aside into cash, which earns nothing here, so losing less than a falling market is the easy bar, and a strategy that is already out clears it automatically. The tables therefore also show who made money and who was in the market and still came out ahead. Windows are peak to trough, known only afterwards, and some are nested (a crash inside a bear market), so events are not independent samples. This is not a forecast.
Latest
Detected automatically
Where each market stands today versus its 60-day high (data through 2026-11-10):
- S&P 500: -3.7% (2026-11-10)
- Gold: -13.7% (2026-11-10)
- Silver: -26.0% (2026-11-10)
- Oil: -16.6% (2026-11-10)
- Natural gas: -9.8% (2026-11-10)
- Commodities: -9.8% (2026-11-10)
- Agriculture: -11.0% (2026-11-10)
- EUR/USD: -4.0% (2026-11-10)
- Bitcoin: -14.9% (2026-11-10)
- Ether: -26.2% (2026-11-10)
Timeline
Newest first. Move = peak to trough (or trough to peak for rallies) on the primary market, measured from daily closes. "Lost less" = share of strategies that did better than the market over the same window.
2026
2025
2024
2023
2022
2021
2020
2019
2018
2017
2016
2015
2014
2013
2012
2011
2010
How the events were measured
- Researched dates are approximate. Each start is snapped to the highest close within 15 trading days of the researched start, and each end to the lowest close within 15 trading days of the researched end, after the start. Rallies use the mirror image (lowest close to highest close).
- A move counts as meaningful only above a threshold: EUR/USD 3%; S&P 500, Gold 5%; Silver, Oil, Natural gas, Commodities, Agriculture 10%; Bitcoin, Ether 20%. Events below it, or on markets with no data yet, are removed (listed below).
- Strategy return = change in the published equity curve from the first to the last day of the window. Before costs, cash earns nothing.
- "Already out" is inferred, not recorded: a strategy counts as in the market if its equity moved on at least one of the last 3 trading days up to and including the window start, using data up to that day only.
- "Lost less than the market" = the strategy's return in the window was higher than the market's. "Next 3 months" = the 63 trading days after the window end.
- Only strategies with a curve on the event's market are replayed; portfolio, basket and multi-asset strategies without one are excluded, so the strategy count differs by market.
- Automatic detection: a market is flagged when its close is at least the threshold below its trailing 60-day high. An episode ends at its trough once price has recovered a third of the fall or 20 trading days have passed since the trough. Episodes detected in the last 60 trading days that are not curated get an automatic page; curated events take precedence.
Researched events we removed
- September 2019 Saudi Aramco attack spike: removed, the measured move of +1.0% on Oil is below the 10% threshold.
- January 2025 DeepSeek sell-off: removed, the measured move of -2.0% on S&P 500 is below the 5% threshold.
- 2010 Flash Crash: removed, it starts at the same peak on the same market as Spring 2010 correction (Greek debt crisis), which covers the whole decline.
The scanner also finds 421 older episodes in the price history that have not been written up; they are not shown.
Markets
S&P 500 · Gold · Silver · Oil · Natural gas · Commodities · Agriculture · EUR/USD · Bitcoin · Ether
Historical, descriptive statistics before trading costs. Not a forecast and not investment advice.