Managing market risk with conditioning information

Excess market return is regressed onto lagged, standardised economic variables. The sign of the forecasted return is used as an input to two market timing models that are run on out-of-sample data during the period 1983–2004. Evidence of market timing ability and superior risk-adjusted performance for the long/short and long-only timing models is demonstrated. Standardised variables appear to have more forecasting power than raw variables. By conditioning on the levels of economic variables with respect to their recent history, market timing leads to better management of market risk and economically significant investment out-performance. Specific applications of Portable Beta are discussed.