Reinforcement learning for option pricing and hedging, a practical edge over black-scholes-merton model
This thesis explores two frameworks leveraging on modern Reinforcement Learning (RL) techniques for pricing and dynamic hedging of an option under practical market conditions such as transaction costs, trader risk-aversion, and stochastic volatility which the Black-Scholes-Merton’s (BSM) model fails...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project (FYP) |
| Language: | English |
| Published: |
Nanyang Technological University
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/175061 |
| Summary: | This thesis explores two frameworks leveraging on modern Reinforcement Learning (RL) techniques for pricing and dynamic hedging of an option under practical market conditions such as transaction costs, trader risk-aversion, and stochastic volatility which the Black-Scholes-Merton’s (BSM) model fails to consider. In this thesis, model-free methods of Q-learning and Deep Reinforcement Learning (DRL) are formulated under these real-world market settings with minimal market assumptions to dynamically hedge in an incomplete market where perfect Delta-hedging through the BSM model is not possible. |
|---|