Learning Pattern Recognition Techniques Applied to Stock Market Forecasting

Most of investment analysis involves decision making by weighing evidence. Such decision processes can be formalized with the aid of pattern recognition (PR) techniques. Specifically, we have applied generalized perceptron-type PR techniques to both general market forecasting and investment selection. And after the investment decision system has been implemented and put into operation, its performance is then gradually improved through learning from previous decision making experiences. Iterative probabilistic learning algorithms (based on stochastic approximation techniques) have been used. Decision models for both investment selection and market forecasting have been realized and tested in actual investment analysis. The experimental results indicate that with the aid of PR techniques we may obtain above average investment performance.