Short-Run Price Efficiency and Discovery of Algorithmic Traders: A Machine Learning Approach

This paper examines the short-run weak-form efficiency of equities, using a proprietary data set permitting analysis of orderbook characteristics to measure short-term price predictability. The results show that high levels of algorithmic trader activity in a stock lowers the level of short-run predictability. We find orderbook imbalance and slope of the order book contain the most information useful for predicting future price movements, and proprietary algorithmic traders lead price discovery visa-vis agency algorithmic traders, highlighting liquidity supplier are not only noise traders.