Volatility autocorrelation in the stock market with artificial neural networks

Abstract

Predicting the trend of financial features in complex financial systems is important and challenging, one useful tool is looking at the autocorrelation function, used in technical analysis as it shows how closely related a pattern reappears in the future. In this paper, we demonstrate a way to optimise the autocorrelation of a linear combination of a stock’s volatility in prices and volumes, lagged at different times using regression neural networks.