Optimal Allocation Across Dark Pools as a Probabilistic Decision Problem

Abstract

A liquidity-conscious trader facing the problem of the efficient execution of a sizable order these days has a multitude of execution venues ranging from traditional exchanges to MFTs to internalizing dark pools and institutional crossing networks. Typically, a more or less heuristic “liquidity seeking algorithm” is employed whereby a randomized strategy repeatedly exposes and re-exposes the order to one or multiple crossing venues while executing some quantity in the lit markets. The efficiency of these heuristic algorithms is never assured and missing is the predictability of the results. There are reasons for relatively slow adoption and development of quantitative dark trading algorithms: dark venues are characterized by a somewhat restrained information outflow. Yet, brokers and traders do possess some quantitative information about the quality of liquidity in the pools. The question remains: is it possible to engage a quantitative theory to make sure that allocation across dark venues is efficient and that it optimally utilizes all of the available information? The author’s answer is: affirmative. In this short article the author presents a straightforward quantitative optimal dark allocation framework. It is based on our experience developing dark allocation algorithms for the EMEA markets.