Stock price prediction method based on cluster of trend vectors

Aiming at the short term price prediction for the investment in stock market, we introduce a way to symbolize the stock price data by data fitting and cluster technology according to its trend, and then propose a predict method using support vector machine which is wrapped by univariate marginal distribution algorithm. Based on the algorithm, we design a corresponding investment strategy, and test it on the real stock data with a comparison to the result of BP neural network. Experimental results show that our method is more effective and performs better than simple BP neural network to guide the investment.