A hybrid model based on stochastic volatility and machine learning to forecast log returns of a risky asset

A hybrid model that combines a stochastic volatility model [2] and the K Nearest Neighbors (KNN) model [1] is proposed to obtain precision forecasts of log returns of a risky asset traded in the financial market. The precision forecasts are the sum of the forecasts obtained with the stochastic volatility model and a correction term produced by the KNN model. Numerical experiments based on real data are performed to investigate the accuracy of the precision forecasts.