Stock price prediction based on Grey Relational Analysis and support vector regression

Abstract

Stock market data is extremely large and complicated. In stock prediction research, the selection of technical indicators has not a scientific theory as a guide. This paper proposes a novel method based on Grey Relational Analysis to select the technical indicators. Then make predictions by Support Vector Regression that optimized by improved fruit fly optimization algorithm. Firstly, the fruit fly optimization algorithm is improved by decreasing footstep and simulated annealing. Secondly, the improved fruit fly optimization algorithm is adopted to optimize the penalty factor c and the kernel function parameter g of the support vector regression. Finally, modeling and forecasting of the stock price with optimized support vector regression are conducted and some simulation experiments are carried out. The Support Vector Regression is adept at analyzing small size and multi-dimensional samples, so it is suitable for short-term stock prediction. By comparing with other three methods, the one this paper proposed could fast convergence and improve the accuracy of forecasting and is an efficient and feasible method.