Empirical Analysis of Limit Order Markets

Abstract

We analyse order placement strategies in a limit order market, using data on the order flow from the Stockholm Stock Exchange. Traders submitting market or limit orders trade off the order price against both the execution probability and the winner's curse risk associated with different order choices. The optimal order strategy is characterized by a monotone function, which maps the liquidity demand of the investors into their order choice. We develop and implement a semiparametric test of this monotonicity property, and find no evidence against the monotonicity property for buy orders or sell orders. We do find evidence against the hypothesis that the trader's decision to be a buyer or a seller depends only on the trading profits available in the limit order book. We estimate that traders submitting market buy orders have private valuations that exceed the asset value by 2.3% on average and receive an average payoff of at least 1.8% of the asset value. Traders submitting limit buy orders at the price below the best ask quote have private valuations between 0.1% and 2.3% above the asset value and earn an average payoff of between 0.3% and 1.8% of the asset value. Although the distribution of liquidity demand does not depend on conditioning information, conditioning information helps us to predict the composition of the order flow in our data. These findings imply that variation in the composition of the order flow can be explained by empirical variation in the relative profitability of alternative order choices and movements in the common value of the asset.

Keywords: Electronic trading systems, limit order markets, semiparametric estimation, auctions

JEL Classification: C14, D44, G10