BEATING THE HOUSE: A STUDY ON IMPROVING BOLLINGER BANDS USING NEURAL NETWORKS

摘要

Much has been written and analyzed when it comes to stock market prediction. Through this discussion, analysts have proven time and time again that linear solutions do not fair well in the prediction of stocks and show little promise that they ever will (Falas et al, 1994). In fact, technical analysts appear to have a leg up when it comes to the debate as they attempt to look for more nonlinear solutions. One of these technical solutions is the use of the Bollinger Band in hopes to better map the volatility of a stock over time (Stridsman, 1997). This band has its flaws, however, and it is the goal of this research to apply a neural network to help aid the band to become more effective in a trending market.