April 21, 2012

Momentum Crashes Revisited

Last December I did a blog post called "Is Momentum Really Dead?" in response to several research papers that purported to show that momentum had lost its edge. Those papers were written due to two unusual occurrences in which long/short stock momentum suffered large loses. The first time this happened was in July/Aug 1932. The second occurrence was in Mar/May 2009. The losses occurred on the short side during sharp market rebounds following periods of extreme, once-in-a-generation, market weakness. I showed that long only momentum and multi-asset momentum never experienced this crash.
Momentum research continues to focus on long/short stock momentum. But long only momentum across asset or risk classes gives better results than long/short momentum using individual stocks. Furthermore, most investors do not care to invest from the short side.

There is a recent paper called Managing the Risk of Momentum by Barroso and Santa-Clara that  adjusts position size according to volatility. This eliminates the crash, substantially reduces drawdown, and gives a boost to the Sharpe ratio.Volatility based position sizing is similar to risk parity and is related to mean variance portfolio sizing.

There are, however, some potential drawbacks to volatility based position sizing in a multi-asset context. First, volatility based portfolios are often heavily concentrated in fixed income. This creates a new risk factor not accounted for in portfolio modeling. With interest rates at such low levels now, the risk of a bear market in bonds is greater than the probability of continued bull market appreciation. On the positive side, interest rates don't often mean revert. They instead tend to drift. There may still be some profits if long bond rates drop further.

The second reason to be cautious of volatility-based position sizing is that lower returns often accompany lower volatility. Although bonds have been a strong performer during the past 15 years, a portfolio heavy in fixed income securities has lower long-run expected returns than a portfolio that is oriented toward higher risk premia assets like equities. One can leverage a lower return fixed income  portfolio to boost returns, but some investors have constraints against the use of leverage. Leverage also increases a portfolio's interest rate, left tail, and correlation convergence risks. Increased leverage may be fine for hedge funds, but it has its drawbacks with respect to other accounts.