Predicting Financial Time Series by Genetic Programming with Trigonometric Functions and High-Order Statistics

This paper describes an extension of the traditional application of Genetic Programming in the domain of the prediction of daily currency exchange rates. In combination with trigonometric operators, we introduce a new set of high-order statistical functions in a unique representation and analyze their performance using daily returns of the British Pound and Japanese Yen. In addition, the same experimental design and analysis is applied to ten other financial time series from two different domains. We will demonstrate that the introduction of high-order statistical functions in combination with trigonometric functions will outperform other traditional models such as ARMA models and Genetic Programming with the basic function set. We utilize the Akaike Information Criterion for the selection of the best ARMA model for our benchmark testing. Performance will be measured on hit percentage, average percentage change, and profit. The t-test is applied to derive confidence intervals and to evaluate the significance of our results. Program Track Evolutionary Computing, Time Series Prediction.