A New Entropic Measure for the Causality of the Financial Time Series

Abstract

A new econometric methodology based on deep learning is proposed for determining the causality of the financial time series. This method is applied to the imbalances in daily transactions in individual stocks and also in exchange-traded funds (ETFs) with a nanosecond time stamp. Based on our method, we conclude that transaction imbalances of ETFs alone are more informative than transaction imbalances in the entire market despite the domination of single-issue stocks in imbalance messages.