Title: Liquidity fluctuations and the latent dynamics of price impact
Authors: Giulia Livieri - Scuola Normale Superiore (Italy) [presenting]
Fabrizio Lillo - University of Bologna and CADS Center for Analysis Decisions and Society Human Technopole Milano (Italy)
Luca Philippe Mertens - Bloomberg LP United States (Italy)
Alberto Ciacci - Blackett Laboratory and Centre for Complexity Science Imperial College London (Italy)
Abstract: Market liquidity is a latent and dynamic variable. Building on the stylized limit order book (LOB) model, we propose a dynamical price impact model at high frequency in which price impact is determined by the product of three components: a daily price impact component, a deterministic intraday pattern, and a stochastic autoregressive component. The resulting model has a linear Gaussian state-space representation which can be estimated using a Kalman filter. We provide empirical evidence in support of our model by analyzing six months of order book data for eight liquid stocks traded on the NASDAQ in 2016. We show that the price change conditional on order flow imbalance predicted by our model explains on average $82\%$ of price change variance. This represents a $16\%$ increase with respect to a previous improved model and a $27\%$ increase with respect to the stylized LOB model. Finally, we perform an out-of-sample analysis of real-time estimates of price impact, and show that our model provides a superior out-of-sample forecast of price impact with respect to historical estimates.