Risk Management of Alpha Models

Abstract

After posting good performance and impressive business growth for over two decades, quantitative equity investment managers have recently produced weak returns. We develop a measure of risk and show how changes in risk provide a common framework to explain past under-performance, as well as improve forecasts of factor returns. We find that quantitative stock ranking models based upon factor weights that vary with their conditional (on risk) forecasted returns are superior to traditional models with fixed weights based upon unconditional historical averages. The suggested improvements to investment processes rely upon objective and well-defined relationships between factor returns and risk, hence justifying our title - risk management of alpha models. Quantitative investors should derive comfort from this research indicating that well-defined modifications to traditional quantitative processes can go a long way towards improving returns and Sharpe ratios as well as in mitigating under-performance.

Keywords: quantitative models, valuation, momentum, conditional forecasting, alpha models, dynamic weighting, multi factor models

JEL Classification: G1, G11, G12, G14