Riding the Waves of Investment Returns

Abstract

It is common knowledge that economic activity follows cyclical processes of different length like the business cycle or seasonal patterns throughout a calendar year. These cyclical patterns are usually distorted by noise and singular events that influence financial markets over the short term. This paper shows how one can filter out the cyclical patterns of investment returns and use this information in order to create superior active investment returns. It is shown that the filtered data will be a good indication of return trends for the asset class considered. In the long run (that is an investment horizon of more than four years) the short term noise should be of negligible importance for the investment performance. This implies that a portfolio that dynamically allocates the portfolio weights according to the expected future filtered returns should be able to add value for long term investors. Here, a simple dynamic portfolio asset allocation process is developed based on the future returns as calculated from the filtered data. The performance of this model portfolio is compared to a passive benchmark portfolio with fixed portfolio weights and maximum Sharpe ratio for the sampling period as well as to an active portfolio that allocates the portfolio weights with a simple momentum approach. The results of our in sample as well as our out of the sample tests show that the model portfolio that relies only on medium to long term cyclical economic activity does outperform both the passive benchmark and the actively managed momentum portfolio in terms of absolute return and in terms of risk adjusted returns (Sharpe ratio). The outperformance is high enough (generally between 1.5% and 5% per annum) to persist even when transaction costs are considered in the model. Also, the Information Ratios of the Model Portfolios are generally in the region between 0.4 and 0.7 for the out of sample data which is an excellent value for a purely quantitative approach. Finally, the superior performance appears to be stable over investment horizons of more than four years.

Keywords: Fourier transformation, investment returns, cycles

JEL Classification: C51, C53, C32