ProtoHedge: interpretable hedging with market prototypes
File(s)
Accepted version
Author(s)
Faloughi, Lisa
Guo, Ce
Luk, Wayne
Type
Conference Paper
Abstract
Deep hedging has emerged as a powerful framework for financial risk management, capable of learning effective hedging strategies in the presence of market frictions. However, its reliance on black-box neural networks creates a critical barrier to adoption, limiting trust, auditability, and regulatory compliance. In this work, we address this challenge by introducing a transparent alternative to the black-box Deep Hedging paradigm. Instead of relying on an opaque architecture, our model learns a finite set of representative market states, or “prototypes”. Hedging decisions are then made via a transparent, similarity-based mechanism: the agent’s action is a weighted average of learned actions associated with each prototype. We call our specific implementation of this framework ProtoHedge. The reasoning approach makes every decision traceable to understandable market scenarios. We conduct extensive experiments in both classical Black-Scholes and more realistic stochastic volatility environments. Our results demonstrate that this interpretability is achieved with minimal impact of less than 0.40% on hedging performance, as our model’s hedging effectiveness is comparable to that of the original black-box deep hedging agent. This work shows that transparency and performance are not mutually exclusive, paving the way for more trustworthy automated risk management systems.
Date Issued
2025-11-14
Date Acceptance
2025-10-10
Citation
ICAIF '25: Proceedings of the 6th ACM International Conference on AI in Finance, 2025, pp.202-210
ISBN
9798400722202
Publisher
ACM
Start Page
202
End Page
210
Journal / Book Title
ICAIF '25: Proceedings of the 6th ACM International Conference on AI in Finance
Copyright Statement
Copyright © 2025 Copyright held by the owner/author(s). This work is licensed under a Creative Commons Attribution International 4.0 License (http://creativecommons.org/licenses/by/4.0/)
License URL
Source
ACM International Conference on AI in Finance
Publication Status
Published
Start Date
2025-11-15
Finish Date
2025-11-18
Coverage Spatial
Singapore