Option market making with hedging-induced market impact
Paulin Aubert Etienne Chevalier Vathana Ly Vath · 2025
Read Option market making with hedging-induced market impact by Paulin Aubert Etienne Chevalier Vathana Ly Vath (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-02518.txt.
Introduction
Paulin Aubert, Etienne Chevalier, and Vathana Ly Vath model a feedback loop that options traders create without noticing. An option market maker quotes prices and then hedges the resulting exposure by trading the underlying asset. When many options change hands, that hedging flow is large enough to move the underlying's price, which in turn changes the options' values. Their model makes option demand a random flow whose intensity depends on the underlying's state and on the maker's own quotes.
The coupled system joins stochastic demand with permanent and transient price impact, so inventory and price evolve together. The authors first examine whether manipulation or arbitrage can arise from the feedback, then prove the mixed control problem, continuous quoting plus impulsive hedging trades, is well-posed. A numerical method based on policy optimisation approximates the optimal strategy and displays how liquidity, inventory risk, and impact trade off.
Catalog record
| CREATOR | Paulin Aubert Etienne Chevalier Vathana Ly Vath |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 101 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-02518
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Option market making with hedging-induced market impact Paulin Aubert Etienne Chevalier Vathana Ly Vath Abstract This paper develops a model for option market making in which the hedging activity of the market maker generates price impact on the underlying asset. The option order flow is modeled by Cox processes, with intensities depending on the state of the underlying and on the market maker’s quoted prices. The resulting dynamics combine stochastic option demand with both permanent and transient impact on the underlying, leading to a coupled evolution of inventory and price. We first study market manipulation and arbitrage phenomena that may arise from the feedback between option trading and underlying impact. We then establish the well-posedness of the mixed control problem, which involves continuous quoting decisions and impulsive hedging actions. Finally, we implement a numerical method based on policy optimization to approximate optimal strategies and illustrate the interplay between option market liquidity, inventory risk, and underlying impact. Keywords: Option market making, Cox processes, Mixed stochastic control, Policy optimization, Machine learning. 1 Introduction Academic research on market making has progressively shifted from stylized microstructure models toward dynamic frameworks able to capture high-frequency trading, inventory risk, and market impact. A key milestone in this evolution is the stochastic-control approach of Avellaneda and Stoikov [ 4 ] , which formulates the quoting problem in continuous time and solves it through Hamilton–Jacobi–Bellman equations. This framework has been extended in various directions, for instance by Guilbaud and Pham [ 12 ] , who incorporated both limit and market orders, or by Guéant et al. [ 11 ] , who derived explicit solutions in an affine setting, enabling practical calibration. Cartea and coauthors [ 8 , 7 ] further broadened the perspective by linking market making to optimal execution and risk management. In parallel, a large literature has developed on optimal execution and market impact, starting with Almgren and Chriss [ 3 ] and enriched by contributions that distinguish temporary and permanent impact [ 10 ] or account for the self-exciting nature of order flow via Hawkes processes [ 2 ] . These advances are highly relevant for option market making, where quoting decisions affect not only order flow but also the underlying dynamics through hedging. Models explicitly integrating such feedback effects have emerged more recently. For example, Stoikov and Saglam [ 16 ] introduced delta-aware option market making, El Aoud and Abergel [ 9 ] modeled joint underlying–option dynamics in a stochastic-control framework, and Baldacci, Bergault and Guéant [ 5 ] developed a multi-option setting where order intensities depend on aggregate sensitivities. While these models provide valuable insight, classical solution techniques become rapidly intractable as soon as one accounts for coupled markets, feedback loops, and state-dependent intensities. This has motivated the use of simulation-based and machine-learning approaches. Spooner et al. [ 15 ] applied reinforcement learning to learn quoting strategies without an explicit parametric model, and Buehler et al. [ 6 ] introduced the Deep Hedging framework, where neural networks optimize strategies directly on simulated paths. In this paper, we consider a market maker mandated to ensure the liquidity of a European option, while complying with hedging constraints imposed by a third party (regulator or principal). Unlike other approaches in the literature, notably [ 16 , 9 , 5 ] , we make no explicit assumptions about the dynamics of the underlying option price. Instead, we model order arrivals in the options market by Cox processes whose intensities depend on both the quotes offered by the market maker and the state of the underlying. These intensities are modeled in a flexible way, allowing the framework to reflect key market effects such as the dependence of order-flow activity on option moneyness and time to expiry. Furthermore, we assume that the market maker has an impact on the price of the underlying via the consumption of the order book, whose shape is sufficiently general to represent low-liquidity assets. This impact is modeled by a permanent and a resilient component. Finally, we develop a numerical method based on neural networks to approximate the market maker’s optimal quoting strategy, while taking into account the associated liquidity and transaction risks. The remainder of the paper is structured as follows. Section 2 introduces the modeling framework and formulates the market-making problem. Section 3 establishes the model’s consistency by examining market integrity and the well-posedness of the control problem. It demonstrates the absence of instantaneous and dynamic arbitrage opportunities, rules out transaction-triggered price manipulation, derives a terminal bound in the coupled setting, and proves quadratic-growth estimates ensuring the finiteness of the value function. Finally, Section 4 develops the numerical methodology and reports experiments. We describe the discrete-time simulator and the neural policy training procedure, then present empirical results in a baseline configuration, two asymmetric scenarios, and a low-liquidity regime. 2 Market modeling and the market maker problem 2.1 Modeling the underlying market Let (\Omega,\mathcal{F},\mathbb{P}) be a probability space equipped with a filtration (\mathcal{F}_{t})_{t\geq 0} satisfying the usual conditions. Throughout this work, all random variables and stochastic processes are defined on (\Omega,\mathcal{F},\mathbb{P}) . 2.1.1 Order book representation In line with the approach of Predoiu, Shaikhet, and Shreve [ 14 ] , we aim to represent the order book of the underlying asset while maintaining full flexibility regarding its structure. In particular, the framework accommodates order books with missing price levels, which may result in a discontinuous shape with gaps. We assume that the shape of the order book is time–invariant. It is characterized by two piecewise continuous functions: f_{B} for the bid side and f_{A} for the ask side. These functions are required to satisfy Assumptions 2.1 and 2.2 . Assumption 2.1 (Integrability of the order book shape) . The order book densities satisfy f_{B}:\mathbb{R}^{-}\to[0,\infty),\qquad f_{A}:\mathbb{R}^{+}\to[0,\infty), together with the integrability conditions: \int_{-\infty}^{0}(1+|u|)f_{B}(u)du<\infty,\qquad\int_{0}^{\infty}(1+u)f_{A}(u)du<\infty. Assumption 2.1 ensures that the total quantity and value represented in the order book are finite. This condition reflects the idea that market participants do not perceive the firm’s fundamental value as unbounded. We also impose that beyond a certain price level no liquidity is available, which is consistent with empirical observations that liquidity does not extend indefinitely. This requirement is formalized in Assumption 2.2 Assumption 2.2 (Right–truncated depth) . There exist U_{A},U_{B}\in(0,\infty) such that f_{B}(u)=0\ \text{for a.e. }u<-U_{B},\qquad f_{A}(u)=0\ \text{for a.e. }u>U_{A}. To further characterize the order book, we introduce cumulative distribution functions that quantify the total volume available when liquidity is consumed up to a given depth. For the bid side, given a current best bid price b>0 , we set \Phi_{B}(x):=\int_{-x}^{0}f_{B}(u)du,\qquad x\in[0,b]. (2.1) The restriction to x\in[0,b] ensures that the bid side does not extend below zero prices, where liquidity is absent. On the ask side we define \Phi_{A}(x):=\int_{0}^{x}f_{A}(u)du,\qquad x\in\mathbb{R}^{+}. (2.2) In this case, no analogous restriction is required, since the ask side can in principle extend arbitrarily far above the current price, with the truncation guaranteed only by Assumption 2.2 . By Assumptions 2.1 and 2.2 , the cumulative functions \Phi_{B} and \Phi_{A} are finite, nondecreasing, and satisfy the bounds \displaystyle\forall b>0,\ \forall x\in[0,b]:\quad 0\leq\Phi_{B}(x)\leq\Phi_{B}(b)\leq\Phi_{B}(U_{B})<\infty, \displaystyle\forall x\in\mathbb{R}^{+}\cup\{+\infty\}:\quad 0\leq\Phi_{A}(x)\leq\Phi_{A}(+\infty)=\Phi_{A}(U_{A})<\infty. The functions \Phi_{B} and \Phi_{A} quantify cumulative liquidity from the best quote up to a given depth: if a market order consumes all liquidity up to level x , the