Forecasting the Stock Price Volatilities by Integratingthe Support Vector Regression and the Krill Herd Algorithm

Abstract

The derivatives which have the property of the high leverage have become popular tools for investing in the era with a low interest rate. Among these derivatives, the options are considered a simpler way for investing since the Black-Scholes (B-S) pricing model can be used to estimate their reasonable prices. However, the parameter "volatility ± in the B-S model cannot be known in advance and needs be guessed based on the historical trading information regarding the options or the underlying assets. Hence, the problems of forecasting future volatilities had become an interesting and attractive research topic for both researchers and practioners. Among the previous researches, the artificial intelligent techniques had been extensively used and acquired satisfactory results. Therefore, the support vector regression (SVR) technique and krill herd (KH) optimization algorithm are utilized to develop an integrated approach for forecasting the volatilities more accurately in this study. The proposed approach is demonstrated by a case study aiming at forecasting the volatilities of the Taiwan Stock Exchange Capitalization Weighted Stock Index (TAIEX) to verify its feasibility and effectiveness. According to the experimental results, the proposed integrated forecasting methodology can produce better forecasting performance, based on the RMSE, R2 or MAPE, than the forecasting models which are built solely based on the SVR. Therefore, it can conclude that the proposed integrated approach can really indeed improve the forecasting, and can be considered an effective and useful assistant tool for an investor to obtain more accurate estimation for the volatility thus helping his/her decision making.