“Machine Learning Techniques for Stock Prediction”

In “Machine Learning Techniques for Stock Prediction”, Vatsal H. Shah (2007) evaluates several machine learning techniques applied to stock market prediction. The techniques used are: support vector machines, linear regression, “prediction using decision stumps”, expert weighting, text data mining, and online learning (the code was from YALE/Weka). The main stock features used were moving averages, exponential moving average, rate of change, and relative strength index. He concludes with “Of all the Algorithms we applied, we saw that only Support Vector Machine combined with Boosting gave us satisfactory results.