Purpose
– Algorithmic trading attempts to reduce trading costs by selecting optimal trade execution and scheduling algorithms. Whilst many common approaches only consider the bid-ask spread when measuring market impact, the authors aim to analyse the detailed limit order book data, which has more informational content.
Design/methodology/approach
– Using data from the London Stock Exchange's electronic SETS platform, the authors transform limit order book compositions into volume-weighted average price curves and accordingly estimate market impact. The regression coefficients of these curves are estimated, and their intraday patterns are revealed using a nonparametric kernel regression model.
Findings
– The authors find that market impact is nonlinear, time-varying, and asymmetric. Inferences drawn from marginal probabilities regarding Granger-causality do not show a significant impact of slope coefficients on the opposite side of the limit order book, thus implying that each side of the market is simultaneously rather than sequentially influenced by prevailing market conditions.
Research limitations/implications
– Results show that intraday seasonality patterns of liquidity may be exploited through trade scheduling algorithms in an attempt to minimise the trading costs associated with large institutional trades.
Originality/value
– The use of the detailed limit order book to reveal intraday patterns in liquidity provision offers better insight into the interactions of market participants. Such valuable information cannot be fully recovered from the traditional transaction data-based approaches.
Liquidity supply and adverse selection in a pure limit order book market
