Delta Hedging of Financial Options Using Reinforcement Learning and an Impossibility Hypothesis

Author

Ronak Tali, Utah State UniversityFollow

Date of Award:

12-2020

Document Type:

Thesis

Degree Name:

Master of Science (MS)

Department:

Mathematics and Statistics

Committee Chair(s)

Kevin R. Moon

Committee

Kevin R. Moon

Committee

Tyler J. Brough

Committee

Jia Zhao

Abstract

In this thesis we take a fresh perspective on delta hedging of financial options as undertaken by market makers. The current industry standard of delta hedging relies on the famous Black Scholes formulation that prescribes continuous time hedging in a way that allows the market maker to remain risk neutral at all times. But the Black Scholes formulation is a deterministic model that comes with several strict assumptions such as zero transaction costs, log normal distribution of the underlying stock prices, etc. In this paper we employ Reinforcement Learning to redesign the delta hedging problem in way that allows us to relax the strict assumption of risk neutrality and allows us to embed market realities such as transaction costs right at the outset. Our main argument is that by taking a controlled amount of risk and encouraging some uncertainty (referred to as exploration) in the hedged position, the market maker is able to generate incremental profit in the entire operation. Our model does not assume any parametric distribution for the underlying stock prices and is fundamentally online in nature i.e. learns on the go.

Checksum

8c1430620ace0c94251b545ebdeb2f12

Comments

All code files are available publicly at https://github.com/ronaktali/MSThesis

Recommended Citation

Tali, Ronak, "Delta Hedging of Financial Options Using Reinforcement Learning and an Impossibility Hypothesis" (2020). All Graduate Theses and Dissertations, Spring 1920 to Summer 2023. 7922.
https://digitalcommons.usu.edu/etd/7922