The three sell-offs

We looked at three periods when US stocks fell sharply, using the same dates for every strategy:

  • Q4 2018: 1 October to 24 December 2018 — SPY −19.2%
  • COVID crash: 19 February to 23 March 2020 — SPY −33.7%
  • 2022 bear market: 3 January to 12 October 2022 — SPY −24.5%

All results are from walk-forward backtests on daily data, before trading costs.

2022 — a slow bear favours the patient

Every strategy lost less than buy-and-hold in 2022, and the median strategy lost 6.8%. Ten finished the period with a gain. The top of the list is dominated by mean reversion: RSI Mean Reversion (+8.9%), Ultimate Oscillator (+8.1%), Z-Score Mean Reversion (+6.3%), Bollinger Band Reversion (+5.7%) and Hurst-Filtered Mean Reversion (+4.1%). Linear Regression Channel (+3.9%) and Crisis Alpha (+3.6%) complete the group.

By style, the median 2022 result was −1.4% for mean reversion, −6.8% for volatility strategies, −7.1% for trend following, −8.4% for regime strategies and −11.4% for momentum. 2022 was a grinding decline with sharp bear-market rallies. Strategies that buy short-term oversold readings and exit quickly caught those rallies; momentum strategies, which need a sustained move, kept getting whipsawed.

The weakest results came from Drawdown Control (−23.8%), Williams %R Reversion (−19.6%), Time-Series Momentum (−18.9%), Parabolic SAR (−17.2%) and 52-Week High Momentum (−16.4%).

2020 — a crash too fast for most rules

The COVID crash took only 23 trading days. The median strategy lost 7.6% while SPY lost a third of its value, but only three strategies made money: Linear Regression Channel (+33.9%), Crisis Alpha (+25.7%) and Connors RSI (+9.4%). Trend strategies did better than their reputation for crashes suggests: the median trend strategy lost 4.7%, and only a quarter lost more than 10%, mostly because they had already stepped aside when the market first wobbled in late February.

Holding up in both

41 of the 62 strategies lost less than half of SPY's decline in both the COVID crash and the 2022 bear market. That sounds reassuring until you look at the other side of the ledger: over the last five years only one strategy beat buy-and-hold SPY on annual return. Protection in sell-offs was paid for with lower returns in the rallies in between.

What to take away

  • No style protects in every kind of sell-off. Fast crashes and slow bear markets reward different behaviour.
  • Mean reversion did well in 2022 because the decline came with violent rallies. In a crash without rebounds it would have kept buying dips.
  • Read crisis results together with the full-period return. A strategy that loses little in a crash but never catches up is not necessarily better.
  • Every strategy page now has a "How it held up in sell-offs" table with these three periods and its returns above and below the 200-day moving average.