Practical Applications of Improving Risk Forecasts Through Cross-Sectional Observations

Abstract

Many traditional approaches to forecasting volatility have shortcomings related to sampling error and other limitations. Jose Menchero , CEO of Menchero Portfolio Analytics Consulting , and Andrei Morozov, Executive Director at MSCI in Berkeley, California, offer a new approach that uses cross-sectional observations to produce more timely and accurate forecasts. This report discusses statistical measures for risk forecasting.

“Volatility forecasting is a challenging task,” Menchero says. “For instance, risk models tend to under-forecast volatility during times of financial crisis, while they over-forecast volatility after the crisis subsides. Our cross-sectional technique mitigates these biases by assigning more weight to recent observations without incurring a high penalty in sampling error.”