Entropy Risk Factor Model of Exchange Rate Prediction

Abstract

We investigate the predictability of an exchange rate with entropy risk factor model, as there is growing evidence that financial markets behave as complex systems. The model is tested on the data of South African Rand (ZAR) exchange rate for the period of 2004-2015. We calculate sample entropy based on the daily data of the exchange rate and conduct empirical implementation of several market timing rules based on these entropy signals. The dynamic investment portfolio based on entropy signals produces better risk adjusted performance than a buy and hold strategy. The returns are estimated on the portfolio values in U.S. dollars. The results raise the potential attractiveness of complex systems analyses, especially the methods of entropy, for foreign exchange market research and applications.