Short Term Trading Models Using Hurst Exponent and Machine Learning

Predicting the direction of Stock Indices has always been an appealing topic which has motivated researchers over the years to develop better predictive models. Recently, Machine learning (ML) based models have been frequently deployed to forecast the direction of classic financial time series data. In the 1950s, Hurst Exponent was introduced as a statistical measure to classify various Time Series. This research analyzes the effectiveness of using Machine Learning and Hurst Exponent along with popular Technical Indicators for short term trading predictions. In this study we explore the use of Hurst Exponent to segment data for a short-term machine learning model in order to improve trading strategy. A comparative analysis has been carried out between the performance of a standalone short-term model, and a Segmented model (Segments based on hurst exponent cut off) in S&P 500, SSE Composite Indices, Gold SPDR Shares and Bitcoin. This new approach is being introduced in order to reach the optimum integration between Machine learning & Hurst Exponent