Gene Expression Programming and trading strategies

Abstract

This paper applies a Gene Expression Programming (GEP) algorithm to the task of forecasting and trading the SPDR Down Jones Industrial Average (DIA), the SPDR S&P 500 (SPY) and the Powershares Qqq Trust Series 1 (QQQ) exchange traded funds (ETFs). The performance of the algorithm is benchmarked with a simple random walk model (RW), a Moving Average Convergence Divergence (MACD) model, a Genetic Programming (GP) algorithm, a Multi-Layer Perceptron (MLP), a Recurrent Neural Network (RNN) and a Gaussian Mixture Neural Network (GM). The forecasting performance of the models is evaluated in terms of statistical and trading efficiency. Three trading strategies are introduced to further improve the trading performance of the GEP algorithm. This paper finds that the GEP model outperforms all other models under consideration. The trading performance of GEP is further enhanced when the trading strategies are applied. © IFIP International Federation for Information Processing 2013.