DO BUSINESS CYCLES EXHIBIT BENEFICIAL INFORMATION FOR PORTFOLIO MANAGEMENT? AN EMPIRICAL APPLICATION OF STATISTICAL ARBITRAGE

An advantageous statistical arbitrage strategy should exhibit a zero-cost tradingstrategy for which the expected payoff should be positive. In practical applications, however,the abnormal returns often are out-of-sample not significant. The statistical modelbeing suggested here results in an estimated portfolio exhibiting in-sample a cointegrationrelationship with the artificial stock index. The portfolio returns exhibited out-of-sample amean of 10.44% p.a., whereas the volatility was one third lower in comparison to the benchmark’svolatility. Accounting for trading costs of 2.94% p.a. on average, the annual returnsof the estimated portfolio are out-of-sample still 6.83% higher than the market returns. Asa result, the model involves implicitly advantageous market timing.