When both the U.S. and World ex-U.S. stock markets are one standard deviation above their mean, the correlation between them is -17%. When both markets are one standard deviation below their mean, the correlation between them rises to +76%.1 This example highlights what many investors have learned during the recent crisis: risk parameters estimated from full samples provide unreliable measures of the hedging and diversification properties of assets during turbulent markets.

We show that it is possible to use multivariate outliers to measure market turbulence. Multivariate outliers coincide with well-known market events. By measuring turbulence, we can estimate risk parameters more reliably and construct portfolios that are more resilient to turbulent markets. Moreover, we can enhance alpha by scaling exposure to risk as a function of market turbulence. A critical feature of our methodology is that it takes into account not only unusually volatile returns but also returns that interact in strange ways. Read the full article here.