Do artificial neural networks provide improved volatility forecasts: Evidence from Asian markets

Abstract This paper enters the ongoing volatility forecasting debate by examining ability of a wide range Machine Learning methods (ML), and specifically Artificial Neural Network (ANN) models. The ANN models are compared against traditional econometric for ten Asian markets using daily data time period from 12 September 1994 to 05 March 2018. empirical results indicate that ML algorithms, across countries, can better approximate dependencies benchmark Notably, predictive performance such deep learning is superior perhaps due its in capturing long-range dependencies. For example, Neuro Fuzzy ANFIS CANFIS, which outperform EGARCH model, more flexible modelling both asymmetry long memory properties. offers new insights markets. In addition standard statistics forecast metrics, we also consider risk management measures including value-at-risk (VaR) average failure rate, Kupiec LR test, Christoffersen independence expected shortfall (ES) dynamic quantile test. study concludes algorithms provide improving forecasts stock Asia suggest this may be fruitful approach management.