Optimal Execution among $N$ Traders with Transient Price Impact

Steven Campbell, Marcel NutzarXiv 2025 · DOI · arXiv

Abstract

We study $N$-player optimal execution games in an Obizhaeva--Wang model of transient price impact. When the game is regularized by an instantaneous cost on the trading rate, a unique equilibrium exists and we derive its closed form. Whereas without regularization, there is no equilibrium. We prove that existence is restored if (and only if) a very particular, time-dependent cost on block trades is added to the model. In that case, the equilibrium is particularly tractable. We show that this equilibrium is the limit of the regularized equilibria as the instantaneous cost parameter $\varepsilon$ tends to zero. Moreover, we explain the seemingly ad-hoc block cost as the limit of the equilibrium instantaneous costs. Notably, in contrast to the single-player problem, the optimal instantaneous costs do not vanish in the limit $\varepsilon\to0$. We use this tractable equilibrium to study the cost of liquidating in the presence of predators and the cost of anarchy. Our results also give a new interpretation to the erratic behaviors previously observed in discrete-time trading games with transient price impact.

Introduction

Institutional investors rarely execute large orders in a single transaction. Doing so would create a massive price dislocation, significantly increasing the cost of the trade. Instead, they employ "optimal execution" strategies, breaking large orders into smaller pieces to be traded over a specific time horizon. The standard mathematical framework for this problem is the Obizhaeva–Wang model, which accounts for "transient price impact"—the phenomenon where a trade moves the market price, but that dislocation decays over time as the market finds a new equilibrium.

While the single-player version of this problem is well-understood, the multi-player extension has historically been a "Wild West" of mathematical finance. In a competitive setting where $N$ traders attempt to liquidate positions simultaneously, a "naive" continuous-time model fails to produce a Nash equilibrium. Traders theoretically attempt to undercut each other infinitely fast, leading to erratic, non-convergent behavior. This paper identifies the missing mathematical components required to restore order to these competitive games.

Methodology

The authors utilize a combination of game theory, stochastic control, and functional analysis to solve the existence problem. The research follows a rigorous four-step process:

  1. Regularization: The game is first "tamed" by adding a small quadratic cost ($\varepsilon$) to the trading rate. This ensures the existence of a unique equilibrium.
  2. Systems of ODEs: The equilibrium is characterized by solving a complex $2N+1$ dimensional system of linear ordinary differential equations.
  3. Closed-Form Solutions: Moving beyond previous numerical approximations, the authors derive fully explicit, closed-form expressions for the equilibrium strategies and associated costs.
  4. Limit Analysis ($\varepsilon \to 0$): The core contribution involves analyzing the behavior as the "speed limit" (the cost $\varepsilon$) is removed. The authors prove that for an equilibrium to exist in this limit, the model must incorporate specific, time-dependent costs on block trades at the start and end of the trading window.

Key Insights

The Necessity of Block Trade Costs

The research identifies that equilibrium existence depends entirely on the choice of block trade costs ($\vartheta_0$ and $\vartheta_T$). For $N > 2$ traders, these costs must be distinct and time-dependent. Specifically, existence is restored if and only if $\vartheta_0 = \lambda(N-1)/2$ and $\vartheta_T = \lambda/2$. This discovery explains why previous discrete-time models exhibited erratic behavior; they were essentially missing these specific market frictions.

Competition and Aggression

In a single-player environment, as trading becomes "cheaper" (friction $\varepsilon \to 0$), the total cost of trading eventually vanishes. However, in an $N$-player game, competition forces traders to be so aggressive that the instantaneous costs converge to a non-zero value. This suggests that market competition creates a floor for execution costs that cannot be removed simply by increasing execution speed.

The Cost of Anarchy and Predatory Trading

The paper quantifies the "Cost of Anarchy"—the efficiency loss incurred because traders compete rather than cooperate. This cost increases with the number of players, eventually plateauing at approximately 12.5% more than a centrally planned execution. Furthermore, the model accounts for "predators": traders who enter the market with zero inventory solely to profit from the price movements caused by others. The presence of even a single predator significantly increases the liquidation costs for other participants.

Conclusion

By deriving a closed-form Nash equilibrium, this paper solves a long-standing problem in trading game theory. It demonstrates that competition fundamentally alters market dynamics, necessitating specific block-trade costs to maintain stability. While the model currently relies on exponential decay kernels and symmetric information, it provides a robust foundation for future research into power-law decay and Mean Field Games ($N \to \infty$).

Summary updated

Abstract

We study $N$-player optimal execution games in an Obizhaeva--Wang model of transient price impact. When the game is regularized by an instantaneous cost on the trading rate, a unique equilibrium exists and we derive its closed form. Whereas without regularization, there is no equilibrium. We prove that existence is restored if (and only if) a very particular, time-dependent cost on block trades is added to the model. In that case, the equilibrium is particularly tractable. We show that this equilibrium is the limit of the regularized equilibria as the instantaneous cost parameter $\varepsilon$ tends to zero. Moreover, we explain the seemingly ad-hoc block cost as the limit of the equilibrium instantaneous costs. Notably, in contrast to the single-player problem, the optimal instantaneous costs do not vanish in the limit $\varepsilon\to0$. We use this tractable equilibrium to study the cost of liquidating in the presence of predators and the cost of anarchy. Our results also give a new interpretation to the erratic behaviors previously observed in discrete-time trading games with transient price impact.
BibTeX
@misc{campbell2025optimal,
  author = {Steven Campbell and Marcel Nutz},
  title = {{Optimal Execution among $N$ Traders with Transient Price Impact}},
  year = {2025},
  eprint = {2501.09638},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2501.09638},
  howpublished = {arXiv},
}
Figure