A Hybrid Forecasting Algorithm Based on SVR and Wavelet Decomposition

Abstract

We present a forecasting algorithm based on support vector regression emphasizing the practical benefits of wavelets for financial time series. We utilize an effective de-noising algorithm based on wavelets feasible under the assumption that a systematic pattern plus random noise generate the data. The learning algorithm focuses solely on the decomposed time series components, leading to a more general approach. Our findings propose how machine learning can be used for data science applications in combination with signal processing methods. Applying the algorithm to real life financial data, we find wavelet decompositions to improve forecasting performance significantly.

Keywords: Support Vector Regression, Forecasting, Algorithm

JEL Classification: C63, C53