Information Horizon, Portfolio Turnover, and Optimal Alpha Models

Abstract

The net value-added of an investment strategy has two parts: the expected value of the alpha skill, and the cost of implementation. The first part is positive and depends on the efficacy of a return forecasting model; the second is negative and depends on portfolio turnover and trading skills. The higher the former, and the lower the latter, the happier the investor. Typically, managers give little consideration to turnover, but throw an alpha model into an optimizer, setting turnover constraints to handle transaction costs. The explicit framework here instead integrates alpha models with portfolio turnover and transaction costs. The result allows evaluation of factors in terms of net return and then optimization of multiple factors to achieve the best net information ratio. Optimal multifactor models change as portfolio turnover constraints change, providing a potential solution for model evolution as assets under management grow.

TOPICS: Portfolio construction, risk management, statistical methods