Date of Award

12-20-2023

Author's School

Graduate School of Arts and Sciences

Author's Department

Statistics

Degree Name

Doctor of Philosophy (PhD)

Degree Type

Dissertation

Abstract

In this thesis, we delve into the intricate optimization challenges of market making, the concurrent provision of buy and sell prices in financial assets. The focus is particularly on the complexities inherent in high-frequency trading scenarios, addressing optimal market making in the presence of latency and incorporating a running inventory penalty. The initial exploration involves the formulation of a stochastic control model that aptly captures the actions of an electronic market maker navigating a trading environment influenced by latency. The main objective of the market maker lies in the maximization of expected terminal wealth. To systematically address and resolve this control problem, we recast it into a finite-horizon Markov Decision Process, subsequently amenable to numerical solutions. A complementary avenue is explored by employing model-free Reinforcement Learning algorithms to tackle the intricacies of market making in the presence of latency. This approach signifies a departure from traditional model-centric methods, harnessing the power of RL to adapt and optimize market-making strategies. We finally propose an extension of the approach introduced in Capponi et al. (2021). This extension introduces a penalty mechanism linked to the running inventory across the entire trading horizon. The incorporation of this running inventory penalty framework significantly enhances the market maker’s risk management capabilities. By doing so, it establishes a more resilient and effective framework for the evaluation of diverse trading strategies. This augmentation is crucial for mitigating the market maker’s exposure to inventory risk, thereby contributing to the robustness of the overall market-making process.

Language

English (en)

Chair and Committee

José Figueroa-López

Download

Share

COinS