FUNDAMENTAL ANALYSIS WITH ARTIFICIAL NEURAL NETWORK

This study performs fundamental analysis and cross-sectional prediction of stock return with neural network technology. Eighteen financial ratios are used as the input vector and one-year ahead stock returns are used as the output vector. The fundamental analysis trading strategy generated by artificial neural networks yields an average annual abnormal return of 22.32% after controlling for market risk, book-to-market, size and momentum effects. Our results highlight neural network’s ability to predict future returns in NYSE/AMEX/Nasdaq securities for the period 1990-2005. Artificial neural network technology stands out as a valuable tool for fundamental analysis and forecasting equity returns in the U.S. markets.