Automatic Method for Stock Trading Strategy

Abstract

A famous hypothesis of price movement, Efficient Market Hypothesis pointed out that due to the uncertainty of companies' news and investors' subjective and personal decisions, the prices of the stock market rely on these factors are actually in a random movement. However, many researchers discovered there are astute investors in the market. They earn money through manipulating the stock trading with an algorithm. Therefore, stock market is not random but systematic. In this project, I first tried different traditional technical analysis method to forecast the stock price, linear regression and regression tree and got a satisfactory result. However, I am looking for a more advanced and scientific way to make trading decision. I kept going on doing research. After that, I learnt to implement an automatic method by combining the technique of machine learning and financial technical analysis to form a forecasting model for Hong Kong's stock trading market. One of the well-known machine learning methods, Support Vector Regression was used to train the stock trend model. Then, a trading strategy, Oscillation Box Theory was introduced to cope with the regression result to optimize the profit. Finally, the results of the trading strategy are always outperforming a buy-and-hold strategy in both bull and bear market.