Title: Finite population games of optimal execution

Abstract: We investigate finite population games of optimal execution, taking place at a market with friction. The models over which we develop our results are akin to the standard Almgren-Chriss model with linear price impacts. On the one hand, at a temporary level, our perspective is rather similar to that of the aforementioned model. On the other hand, all players in the model will impact the asset's public price, yielding an aggregate permanent price impact. We propose to analyze two different settings. The first one comprises the case where there is no hierarchy among players, and there is a symmetry of information. In this setting, we obtain closed-form formulas to the Nash equilibrium in the most general setting, i.e., when players' preferences are completely heterogeneous. Particularizing to the case of homogeneous parameters, we show that the average optimal inventory of the finite population converges to its mean-field counterpart, uniformly over a fixed trading horizon, as the population size grows to infinity. In the second framework, we consider a major player, also called a leader, with the first move advantage, and a population of minor players, also known as followers, thought of as high-frequency traders, which trade on informational advantage against the leader. This leads to a model of McKean-Vlasov type for the dynamics of the asset's midprice. We prove the existence and uniqueness of the Stackelberg-Nash equilibrium for a reasonable set of model parameters. We also characterize it as the solution of an abstract vector forward-backward stochastic differential equation system.