Mark Kritzman is rewriting conventional wisdom about risk and diversification. His concept of “turbulence,” a statistical measure of volatility and correlation, allows managers to construct portfolios that are more resilient in today’s markets conditions.

Traditionally, advisors have used software to navigate the efficient frontier, based on estimates of risk (standard deviation) and return by asset class.  New research shows that using turbulence can produce portfolios with better risk-return profiles.

Mark Kritzman, CEO of Cambridge, MA-based Windham Capital Management, presented his research at the FPA annual conference in Boston last month.  Kritzman also teaches a course in financial engineering at MIT and serves on the editorial boards of a number of academic publications.

Advisors should construct portfolios differently, Kritzman said at the conference, depending on whether markets are in a quiet or turbulent period.  Turbulence is measured by looking at price changes and at correlations between asset classes.  Turbulence occurs when these two variables behave in ways that are significantly different from their historical patterns.  Conventional measures of volatility, such as the VIX Index (which measures the volatility of the S&P 500), are limited because they measure only one asset class (e.g., equities) and do not look at correlations across asset classes.

The current credit crisis has been accompanied by record levels of turbulence, regardless of the set of assets measured.

Turbulence levels now rival those seen in the US markets during the tech bubble, and currency market turbulence is now reminiscent of the Russian default.  In the global markets, current levels of turbulence are unprecedented.

The data above is a 30-day moving average of the index.  Turbulence in the US markets peaked on September 29, the day the first bailout bill was defeated, registering a value of 147.24.

When turbulence arrives may be random, but once it arrives it persists.  For example, the following data shows the turbulence index for US markets:

These data show how long turbulence lasts, based on whether the turbulence index falls outside a certain threshold.  For example, in the US markets, if the turbulence index falls in the top 10% of its historical sample (equating to a turbulence index of 15.99), then for the week following this event the average turbulence (22.66) remains high.  Even two weeks or a month after the onset turbulence has not subsided.  The importance, for investors, is that they have time to react to turbulence once it arrives.

Display article as PDF for printing.

Would you like to send this article to a friend?