How to Hedge an Option Against an Adversary: Black-Scholes Pricing is Minimax Optimal

Jacob D. Abernethy, Peter L. Bartlett, Rafael M. Frongillo, Andre WibisonoNeurIPS 2013 · OpenReview

Authors: Jacob Abernethy, Peter L. Bartlett, Rafael M. Frongillo, and Andre Wibisono

The Black-Scholes model has been the bedrock of option pricing since its introduction in 1973. However, its foundational assumption—that asset prices follow Geometric Brownian Motion (GBM)—has long been a target for criticism. In a GBM world, prices move continuously and stochastically, whereas real-world markets are messy, exhibiting sudden price jumps, trading halts, and non-random behaviors. This paper bridges classical financial mathematics and modern online learning theory to provide a robust, non-stochastic justification for the world's most famous financial formula.

The authors demonstrate that the classic Black-Scholes option pricing model is "minimax optimal." Even if the stock market behaves like a malicious adversary actively trying to maximize an investor's losses, the Black-Scholes strategy remains the optimal hedging approach, provided the total market volatility remains bounded.

The Adversarial Market Framework

Instead of relying on stochastic processes, the authors reframe option hedging as a sequential zero-sum game between Nature (the market) and an Investor. In this minimax regret framework:

  • The Investor decides the allocation of capital in an asset at each time step.
  • Nature determines the subsequent price fluctuation of that asset.

A core conceptual leap in this paper is the treatment of volatility. In standard finance, volatility is a statistical parameter. Here, it is reimagined as a "budget" or resource that the adversary spends to manipulate the price. Nature is restricted by two primary constraints:

  1. Total Volatility Constraint: The cumulative quadratic volatility (the total budget for price fluctuations) is capped at a constant $c$.
  2. Jump Constraint: Individual price jumps cannot exceed a specified maximum size $\zeta$.

Hedging Strategy and Convergence

To analyze the investor's "regret"—the difference between the hedging strategy's performance and the actual payoff of the option—the authors employ a combination of Taylor expansions and Partial Differential Equations (PDEs).

The paper provides an explicit hedging strategy for the investor based on a modified version of "Delta Hedging," where shares are bought or sold based on the derivative of the option's value. The authors prove that as the number of trading rounds increases and the maximum jump size ($\zeta$) decreases, the value of this adversarial game converges exactly to the Black-Scholes price, $U(S, c)$.

Unlike previous attempts to link minimax games to Black-Scholes, this work successfully incorporates discrete price jumps. The authors establish a concrete upper bound on the investor's loss: $V \le U + O(c \zeta^{1/4})$. This formula quantifies the maximum "extra" cost incurred by hedging against an adversarial market compared to a standard random market, showing that the error term vanishes at a rate of $O(\zeta^{1/4})$.

Implications and Limitations

The most striking insight from this research is the profound robustness of the Black-Scholes formula. It emerges naturally as the solution to a minimax game, suggesting the model is highly effective not just for "random" markets, but for any market dynamic that stays within a total volatility budget.

However, the theoretical framework does have limitations:

  • Known Volatility: The model assumes the investor knows the total volatility budget $c$ in advance. In practice, estimating future volatility remains one of the most challenging aspects of quantitative trading.
  • Payoff Constraints: The mathematical proofs require the option's payoff function to be convex and "eventually linear." While this perfectly describes standard European Call and Put options, it may not generalize to more exotic financial derivatives.
  • Tightness of the Bound: The authors note that the $O(\zeta^{1/4})$ error bound is likely not perfectly tight. Future mathematical refinements could potentially improve this bound to $O(\zeta^{1/2})$ or even $O(\zeta)$.

Ultimately, this paper provides a compelling new lens through which to view option pricing, proving that the Black-Scholes model's utility extends far beyond the restrictive assumptions of Geometric Brownian Motion.

Summary updated

Abstract

We consider a popular problem in finance, option pricing, through the lens of an online learning game between Nature and an Investor. In the Black-Scholes option pricing model from 1973, the Investor can continuously hedge the risk of an option by trading the underlying asset, assuming that the asset's price fluctuates according to Geometric Brownian Motion (GBM). We consider a worst-case model, in which Nature chooses a sequence of price fluctuations under a cumulative quadratic volatility constraint, and the Investor can make a sequence of hedging decisions. Our main result is to show that the value of our proposed game, which is the regret'' of hedging strategy, converges to the Black-Scholes option price. We use significantly weaker assumptions than previous work---for instance, we allow large jumps in the asset price---and show that the Black-Scholes hedging strategy is near-optimal for the Investor even in this non-stochastic framework."
BibTeX
@inproceedings{abernethy2013how,
  author = {Jacob D. Abernethy and Peter L. Bartlett and Rafael M. Frongillo and Andre Wibisono},
  title = {{How to Hedge an Option Against an Adversary: Black-Scholes Pricing is Minimax Optimal}},
  booktitle = {Advances in Neural Information Processing Systems},
  year = {2013},
  url = {https://openreview.net/forum?id=GuRKiKGyIG},
}
Figure