Order Flow Decomposition for Price Impact Analysis in Equity Limit Order Books

The advent of electronic trading sparked the development of an extensive body of literature dedicated to the study of market microstructure. A key objective in this field is to understand the determinants of price changes, by considering a wide range of variables based on the very rich information provided by the Limit Order Book (LOB). Previously, it has been shown that short term price movements are dictated by the order flows, and, specifically, by the Order Flow Imbalance (OFI). In this work, we study the price impact of order flows, in both contemporaneous and forward-looking scenarios. By building on state-of-the-art results in the field, and by leveraging the potential of market-by-order data, we introduce the Decomposed OFI, a novel extension of the standard OFI from the literature, which incorporates knowledge about order book event types, in addition to order flow information. Our extensive set of experiments, based on 100 stocks for three years of market-by-order data, reveal that our proposed Decomposed OFI models lead to a significant improvement in a forward-looking predictive scenario, from both a statistical and an economic benefits point of view. Moreover, in a contemporaneous prediction setting, models based on Decomposed OFI yield comparable results to standard OFI-based models. Our work opens up further avenues of research in terms of conditional order flow imbalances in equity limit order books.