Reinforcement Learning for Option Hedging: Static Implied-Volatility Fit versus Shortfall-Aware Performance

TL;DR

This paper introduces new reinforcement learning methods for option hedging that incorporate market frictions and risk preferences, demonstrating improved static implied-volatility fit and dynamic hedging performance on real market data.

Contribution

It extends existing RL-based option pricing models with risk aversion and trading costs, and proposes a novel RLOP approach for better hedging outcomes.

Findings

Adaptive-QLBS improves static implied volatility fit.

RLOP enhances dynamic hedging by reducing shortfall risk.

Both methods outperform traditional models in their respective metrics.

Abstract

We extend the Q-learner in Black-Scholes (QLBS) framework by incorporating risk aversion and trading costs, and propose a novel Replication Learning of Option Pricing (RLOP) approach. Both methods are fully compatible with standard reinforcement learning algorithms and operate under market frictions. Using SPY and XOP option data, we evaluate performance along static and dynamic dimensions. Adaptive-QLBS achieves higher static pricing accuracy in implied volatility space, while RLOP delivers superior dynamic hedging performance by reducing shortfall probability. These results highlight the importance of evaluating option pricing models beyond static fit, emphasizing realized hedging outcomes.