Do equity strategy portfolios characterized by aggressive (conservative) scaling when portfolio volatility is recently low (high) reliably beat unmanaged performance? In their March 2019 paper entitled “On the Performance of Volatility-Managed Portfolios”, Scott Cederburg, Michael O’Doherty, Feifei Wang and Xuemin Yan assess whether practical volatility management is systematically attractive. For each of 103 anomalies (nine widely used factors and 94 other published anomalies), they construct a hedge portfolio that is each month long (short) the value-weighted tenth of stocks with the highest (lowest) expected returns. They then construct volatility-managed versions of these portfolios based on inverse variance of daily portfolio returns the prior month. Focusing on gross Sharpe ratio, they compare head-to-head performances of volatility-managed portfolios and unmanaged counterparts. Focusing on gross Sharpe ratio and certainty equivalent return (CER), they also employ an historical training subsample to estimate mean-variance optimal allocations for: (1) a strategy that chooses among a given volatility-managed portfolio, its unmanaged counterpart and a risk-free asset; and, (2) a strategy chooses between only the unmanaged counterpart and the risk-free asset. Using daily returns for the 103 equity hedge portfolios, they find that:
- In simple head-to-head competitions, volatility-managed portfolios have higher gross Sharpe ratios than unmanaged counterparts for only 53 of 103 strategies. Just eight of 53 winning margins are statistically significant.
- In-sample regressions of volatility-managed portfolio returns versus unmanaged counterpart returns, scaled so that their full-sample portfolio volatilities are equal, generate positive alphas for 77 of 103 strategies. 23 (three) volatility-scaled alphas are significantly positive (negative). However, scaling to equalize full-sample volatilities unrealistically requires full-sample foresight.
- Reasonable out-of-sample mean-variance optimal implementations produce lower gross Sharpe ratios and CERs for 72 of 103 strategies when including a volatility-managed choice than when not including this choice. An array of robustness checks confirms this poor out-of-sample performance.
In summary, evidence suggests investors should be skeptical about practical value of volatility-managed equity hedge portfolios as claimed in prior studies.
Cautions regarding findings include:
- Strategy returns are gross, not net. Accounting for monthly hedge portfolio reformation frictions and shorting costs/constraints would lower returns. Costs may differ for volatility-managed portfolios and unmanaged counterparts, such that net findings may differ from gross findings.
- Volatility management of diversified stock portfolios is beyond the reach of many investors, who would bear fees for delegating to a fund manager.
- Testing many strategies on the same sample introduces data snooping bias, such that best-performing strategies overstate expectations.