Fast learning and predicting of stock returns with virtual generalized random access memory weightless neural networks
Stock market prediction focus on developing approaches to determine the future price of a stock or other financial product. The key task of stock market prediction is to determine the timing for the buying or selling of stock, undoubtedly, it is very difficult due to the high volatility and nonlinear relationships driven by short-term fluctuations in investment demand. In this work, we address this problem by adopting the Cox's hazard model to predict a stock's future rising or dropping probabilities. Specifically, we define the problem of Buy-and-Sell-Point Prediction from the survival analysis perspective. The Cox's hazard model is proposed as the model of choice for this prediction problem due to several reasons including the ability to model the dynamics in the stock movement, and to easily incorporate different types of technical indexes as covariates. In the experiment, we apply the trained model for the stock market forecasting on six stocks in Shanghai Stock Exchange. The results show that the proposed model is superior to several baseline models in terms of accuracy, and the stock return evaluations have revealed that the profits produced by the proposed model are higher.