In practice, there exist three different "religions" about stock market prediction. First of all, there is the Random Walk hypothesis, which basically says that the stock market is not predictable at all. In the second theory - the Fundamental Analysis - forecasts are based on economic data. Finally, there is the Technical Analysis approach, which is trying to predict future price changes using
... [Show full abstract] only historical prices and volumes. In this work, we are studying the third approach. We consider twenty different models based on various technical data. These models depend also on a small number of parameters. In our prediction process, we try to maximize our gain by making one decision every day - either buy or sell. To make the decision at a certain day, we use a model whose parameters are optimized to yield a maximum gain over a particular "training time" - the most recent 5, 10, 25, 50, 75 or 125 past days. We test the models and prediction process with various stocks over a simulation t...