Martingale Optimal Transport for Model-Independent Finance

You are reading version 1 of this direction, frozen 2026-10-04. The latest version is version 2.

Version 1 of 2118 papers251 conceptsfrozen 2026-10-04

In classical financial mathematics, pricing a derivative usually begins by assuming a specific model for asset prices, such as a random walk or a diffusion process. However, if the model is wrong, the price is wrong—a vulnerability known as model risk. Martingale Optimal Transport (MOT) offers a "model-independent" alternative. Instead of picking one model, MOT considers the entire universe of possible models that are consistent with observed market data and identifies the absolute highest and lowest possible prices for a derivative.

The Martingale Constraint: A Rigid Geometry

Standard Optimal Transport (OT) asks for the most efficient way to move mass from one distribution $\mu$ to another $\nu$. In finance, these distributions represent asset prices at different points in time, often inferred from the prices of vanilla call options. However, standard OT is "blind" to the fundamental requirement of efficient markets: the no-arbitrage condition.

In a fair market, the current price of an asset must be the best predictor of its future price. This is the martingale property. As noted by Mathias Beiglboeck, Jan Obloj, and Tongseok Lim in Dual attainment for the martingale transport problem, this constraint adds significant "structural rigidity." While standard OT allows mass to be moved anywhere, MOT requires that for every starting point $x$, the average of where that mass ends up must be $x$ itself.

This constraint transforms the problem from a simple cost-minimization into a generalized moment problem. As explored in A Harmonic Analysis Solution to the Static Basket Arbitrage Problem, observed market prices act as constraints on an underlying probability measure. The goal is to find the "worst-case" measure that extremizes the price of a complex derivative while remaining a martingale.

The Duality of Pricing and Hedging

The power of MOT lies in its mathematical duality, which bridges the gap between two different financial perspectives:

  1. The Primal Problem (Pricing): Finding the martingale measure that maximizes the expected payoff of a derivative.
  2. The Dual Problem (Hedging): Finding the cheapest "superhedging" strategy—a portfolio of traded assets that is guaranteed to pay out at least as much as the derivative, regardless of what the market does.

Marcel Nutz, Florian Stebegg, and Xiaowei Tan demonstrate in Multiperiod Martingale Transport that "strong duality" holds: the maximum price from the transport problem is exactly equal to the minimum cost of a robust hedge. This equivalence ensures that the price bounds are not just theoretical abstractions but correspond to executable trading strategies.

In this dual framework, the Lagrange multipliers used to solve the optimization problem have a physical meaning: they represent the weights of vanilla options in a static hedge. As Pierre Henry‐Labordère and Nizar Touzi show in The maximum maximum of a martingale with given n marginals, these strategies often involve "semi-static" hedging—combining a fixed portfolio of options with dynamic trading of the underlying asset triggered by specific price barriers.

Stability and Robustness in Noisy Markets

For MOT to be useful in practice, it must be stable. If a small change in market prices led to a massive change in the calculated price bounds, the framework would be too fragile for real-world use.

Recent work by Gudmund Pammer and Julio Backhoff‐Veraguas in Stability of martingale optimal transport and weak optimal transport proves that MOT is indeed stable. They show that small perturbations in marginal distributions (the input data) result in only small changes to the optimal transport plan. This stability is achieved by using the "adapted" version of the weak topology, which respects the temporal flow of information—a crucial detail for stochastic processes.

Furthermore, Ariel Neufeld and Marcel Nutz extended these results to handle "rough" payoffs, such as digital options, in Superreplication under Volatility Uncertainty for Measurable Claims. They established that the duality between the worst-case expectation and the superreplication price remains valid even for irregular, non-smooth financial instruments.

Continuous Time and the Skorokhod Connection

While many MOT models operate in discrete time steps, the field has expanded into continuous-time paths. This connects MOT to the Skorokhod Embedding Problem (SEP): finding a stopping time for a Brownian motion such that its position follows a target distribution.

In Fine Properties of the Optimal Skorokhod Embedding Problem, researchers like Mathias Beiglböck show that solving the SEP is essentially a continuous-time version of MOT. The "optimal" way to stop the process often occurs the moment it hits a specific geometric barrier. This geometric intuition is further refined by Nassif Ghoussoub and Young-Heon Kim in A solution to the Monge transport problem for Brownian martingales, where they prove that the optimal martingale transport is unique and follows a rigid geometric structure in arbitrary dimensions.

By framing derivative pricing as a transport problem, researchers have created a unified language—bridging optimal control, PDE theory (like the Hamilton-Jacobi-Bellman equations), and probability—to ensure that financial risks are managed not just against a single model, but against the market's inherent uncertainty.

Papers in this version

Showing 60 of 118 papers.

History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 98 papers
  2. Version 1 · 2026-10-04 · 118 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_pricing_hedging_duality_b321528b,
  title = {pricing-hedging duality},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/pricing-hedging-duality/rdr_b321528bdf4d4306addb82aa66278a91},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_martingale_optimal_transport_for_model_i_b321528b_v1,
  title = {Martingale Optimal Transport for Model-Independent Finance},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/martingale-optimal-transport-for-model-independent-finance/rdr_b321528bdf4d4306addb82aa66278a91/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}