How well did previously identified portfolio risk management strategies work during the COVID-19 market crash? In their July 2020 paper entitled “Strategic Risk Management: Out-of-Sample Evidence from the COVID-19 Equity Selloff”, Campbell Harvey, Edward Hoyle, Sandy Rattray and Otto Van Hemert extended analyses of risk management strategies they identified in a 2016-2019 series of papers with an out-of-sample test of the February-March 2020 stock market sell-off. These strategies include:
- Long put options, short credit risk, long bonds or long gold.
- Trend following based on time series/intrinsic momentum (past return divided by volatility of returns over a specified lookback interval) or on moving average crossovers.
- Holding defensive stocks (based on profitability, payout, growth, safety or quality).
- Volatility targeting (increasing/decreasing exposure when past volatility is relative low/high).
- Rebalancing a stocks-bonds portfolio only half way and only when recent (1, 3 or 12 months) portfolio return is above its historical average.
Extending analyses from their prior papers through March 2020 to capture the COVID-19 crash, they find that:
- Buying puts during the sharp COVID-19 sell-off worked well, but short credit risk worked even better (however, these strategies do not work well across all market conditions.) Neither long bonds nor long gold worked well.
- Fast (1-month) trend following performed well during the COVID-19 sell-off, and 3-month momentum worked reasonably well. Time series momentum as specified worked better than comparable moving average crossovers. (Trend following works reasonably well across all market conditions.)
- Long-short portfolios based on profitability or growth worked reasonably well during the COVID-19 sell-off, but that based on safety or quality did not.
- Volatility targeting dramatically reduced equity allocation ahead of the highest volatility in March 2020, thereby improving return and risk at that time.
- Strategic stocks-bonds rebalancing rules helpfully kept equities underweighted at the end of February 2020.
In summary, evidence suggests that trend following strategies offer reasonable protection from crashes such as the COVID-19 sell-off with reasonable performance across other market conditions.
Cautions regarding findings include:
- Findings are gross, not net. Some of the above strategies may be more difficult and costly to implement (such as diversified long-short stock portfolios) than others.
- Some strategies problematically imply precise market timing (holding puts during a crisis, or shifting from a slower to faster momentum strategy). Others are more readily manageable across market conditions.
- Testing multiple strategies on the same data introduces data snooping bias, such that the best-performing strategy overstates expectations.
- Further extending tests of the risk management strategies through the sharp rebound after March 2020 may be instructive.
See also “U.S. Stock Market Crisis Hedge Strategies”, “Benefits of Volatility Targeting Across Asset Classes” and “Best U.S. Equity Market Hedge Strategy?”.
Frequently Asked Questions
What is volatility targeting in portfolio risk management?
Volatility targeting is a risk management strategy that increases or decreases portfolio exposure when past volatility is relatively low or high, respectively. In the Harvey et al. out-of-sample test of the February-March 2020 COVID-19 sell-off, volatility targeting dramatically reduced equity allocation ahead of the highest volatility in March 2020, improving return and risk at that time.
Does trend following protect a portfolio during a stock market crash?
Trend following offered reasonable protection during the February-March 2020 COVID-19 sell-off, according to Harvey et al. Fast (1-month) trend following performed well and 3-month momentum worked reasonably well. Trend following also performed reasonably across other market conditions, though results are gross of costs and subject to data snooping bias.
How is time series momentum measured for trend following?
Time series (intrinsic) momentum is measured as past return divided by the volatility of returns over a specified lookback interval. Harvey et al. compared this approach with moving average crossovers using different lookback horizons, such as 1-month and 3-month. In their COVID-19 sell-off test, time series momentum as specified worked better than comparable moving average crossovers.
How do long bonds and gold compare with put options and short credit risk as crash hedges?
In the Harvey et al. test of the February-March 2020 COVID-19 sell-off, neither long bonds nor long gold worked well as hedges. Buying put options worked well, and short credit risk worked even better. However, puts and short credit risk do not work well across all market conditions.
What are the limitations of using risk management strategies like puts and trend following to hedge crashes?
The Harvey et al. findings are gross rather than net of costs, and some strategies, such as diversified long-short stock portfolios, may be difficult and costly to implement. Some strategies imply precise market timing, such as holding puts during a crisis or shifting from slower to faster momentum. Testing multiple strategies on the same data also introduces data snooping bias, so the best-performing strategy overstates expectations.