Paper: arXiv 2611.26119
Authors: Mohamad Hassan Abou Daya, Carole Bernard, Steven Vanduffel
Abstract
We construct sub-replicating portfolios of vanilla options for a basket call when the marginal distributions of the underlying assets are known and their dependence is not. To do so, we reduce the n-asset problem to n two-asset problems of the type solved by Hobson, Laurence, and Wang (2005a), and use the Block Rearrangement Algorithm1 (BRA) to approximate the minimising dependence among the assets. The cash component of the portfolio is computed exactly, so that its price is a valid lower bound, and reassembling the BRA output with the original marginals gives an upper bound computed by quadrature; together they bracket the minimum basket-option price. In log-normal examples with three and six assets the bracket has a relative width of at most 0.1%, and the corresponding sub-replicating portfolios are sparse, using only two strikes per asset, while the resulting bounds are tighter than those obtained from a general-purpose multi-marginal optimal-transport algorithm with strikes on a grid. We also give the lower bound and an optimal portfolio in closed form in three special cases: a deep in-the-money basket, a basket with one dominant asset, and jointly mixable marginals.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: This paper presents a highly mathematical approach to a complex problem in quantitative finance, demonstrating strong empirical results through numerical examples. The novelty lies in its efficient use of the Block Rearrangement Algorithm and a pair decomposition to construct sparse sub-replicating portfolios, offering tighter bounds than existing methods.
Research Flowchart
flowchart TD
A["Research Goal: Construct Sub-Replicating Portfolios for Basket Options"] --> B{"Key Methodology: Reduce n-asset to n 2-asset problems & use BRA"};
B --> C["Inputs: Known Marginal Distributions, Basket Option Parameters"];
C --> D["Computational Processes: Block Rearrangement Algorithm (BRA), Quadrature for Upper Bound, Closed-Form for Cash Component"];
D --> E{"Key Findings: Sparse Portfolios (2 strikes/asset), Tight Bounds (relative width <= 0.1% for log-normal examples)"};
E --> F["Outcomes: Valid Lower Bound, Upper Bound, Tighter than General Multi-Marginal OT"];
F --> G["Special Cases: Closed-form for deep ITM, dominant asset, jointly mixable marginals"];
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper proposes a method to construct feasible sub‑replicating portfolios of vanilla options for basket calls by reducing the n‑asset problem to n two‑asset problems and using the Block Rearrangement Algorithm to approximate the minimizing dependence, with an exact cash component and a quadrature
Source sentence
We construct sub-replicating portfolios of vanilla options for a basket call when the marginal distributions of the underlying assets are known and their dependence is not. To do so, we reduce the n-asset problem to n two-asset problems of the type solved by Hobson, Laurence, and Wang (2005a), and use the Block Rearrangement Algorithm1 (BRA) to approximate the minimising dependence among the assets.
Assumes 3
- All analysis is performed under a risk‑neutral measure with zero interest rates, no dividends, and unit weights in the basket. market
Source sentence
Throughout the paper we work under a risk-neutral measure Q, assume zero interest rates, no dividends, and unit weights in the basket; the extension to non-unit weights and constant interest rates follows by a trivial scaling argument (Laurence and Wang 2005).
- Each marginal distribution of the underlying assets admits a density on (0,∞) with full support and finite mean. model
Source sentence
For each i = 1, . . . , n, the marginal law of Xi under Q admits a density on (0, ∞) with full support and finite mean.
- The call price function for each asset is strictly positive, strictly decreasing, strictly convex and continuously differentiable. model
Source sentence
The call price function CXi(k) is therefore strictly positive for every k ⩾0, strictly decreasing, strictly convex on [0, ∞), and continuously dif-ferentiable in k, with derivative ∂kCXi(k) = −Q(Xi > k) = −Q(Xi ⩾k) (the two coincide because {Xi = k} has measure zero under the density). Remark 2.2.
Data
- log‑normal marginal distributions synthetic unknown
Source sentence
In log-normal examples with three and six assets the bracket has a relative width of at most 0.1%, and the corresponding sub-replicating portfolios are sparse, using only two strikes per asset, while the resulting bounds are tighter than those obtained from a general-purpose multi-marginal optimal-transport algorithm with strikes on a grid.
Costs
Not applicable
Finds 4
- In log‑normal examples with three and six assets the lower‑upper bound bracket has a relative width of at most 0.1% (and 0.02% for six assets). theoretical figure from elsewhere in the paper
Source sentence
In log-normal examples with three and six assets the bracket has a relative width of at most 0.1%, and the corresponding sub-replicating portfolios are sparse, using only two strikes per asset, while the resulting bounds are tighter than those obtained from a general-purpose multi-marginal optimal-transport algorithm with strikes on a grid. The price of the portfolio under the original marginals is a lower bound,
- Using only two strikes per asset, the sub‑replicating portfolios achieve higher lower bounds than grid‑based methods that use 32 to 256 strikes per asset. performance
Source sentence
With two strikes per asset our lower bounds are higher than theirs with 32 to 256 grid strikes per asset and our upper bounds are considerably tighter.
- Adding additional grid strikes to the two‑strike portfolios does not improve the lower bound, indicating strike location is crucial. theoretical
Source sentence
Moreover, adding grid strikes to ours does not improve our lower bound. This shows that the location of the strikes, and not their number, is crucial for obtaining tight bounds.
- The BRA‑based dependence structure and the resulting portfolio are optimal for the original marginals to within 10⁻¹¹ quadrature accuracy. theoretical
Source sentence
Our bracket closes: the two bounds differ by less than 10−11, which is the accuracy of the quadrature, so the BRA dependence structure and the portfolio are optimal for the original marginals to that accuracy.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 18 related papers). Coverage is partial, so this is neither replication nor refutation.
Extracted 2026-11-20 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (32)
- Static-arbitrage optimal subreplicating strategies for basket options (2005)
- Sharp upper and lower bounds for basket options (2005)
- Static-arbitrage lower bounds on the prices of basket options via linear programming (2010)
- Static-arbitrage upper bounds for the prices of basket options (2005)
- Model-free bounds for multi-asset options using option-implied information and their exact computation (2023)
- Numerical method for feasible and approximately optimal solutions of multi-marginal optimal transport beyond discrete measures (2022) — link
- Computing lower bounds on basket option prices by discretizing semi-infinite linear programming (2016)
- Rearrangement algorithm and maximum entropy (2018)
and 24 more
- Algorithms for finding copulas minimizing convex functions of sums (2016)
- Block rearranging elements within matrix columns to minimize the variability of the row sums (2018)
- Improved block rearrangement algorithm (2026)
- Computation of sharp bounds on the distribution of a function of dependent risks (2012)
- Risk aggregation with dependence uncertainty (2014)
- Joint mixability (2016)
- Static arbitrage bounds on basket option prices (2006)
- Computing general static-arbitrage bounds for European basket options via Dantzig-Wolfe decomposition (2010)
- A novel feasible discretization method for linear semi-infinite programming applied to basket option pricing (2011)
- Dependency of risks and stop-loss order (1996)
- Model-independent bounds for option prices—a mass transport approach (2013)
- Bounds on multi-asset derivatives via neural networks (2020)
- Improved model-free bounds for multi-asset options using option-implied information and deep learning (2024) — link
- Robust risk aggregation with neural networks (2020)
- Robust pricing and hedging of options on multiple assets and its numerics (2021)
- Marginal and dependence uncertainty: bounds, optimal transport, and sharpness (2022)
- Improved robust price bounds for multi-asset derivatives under market-implied dependence information (2024)
- Duality theorems for marginal problems (1984)
- Inequalities for distributions with given marginals (1980)
- Solution of a statistical optimization problem by rearrangement methods (1983)
- An Introduction to Copulas (2006)
- A simplex method for function minimization (1965)
- An efficient method for finding the minimum of a function of several variables without calculating derivatives (1964)
- On the distributional transform, S klar’s theorem, and the empirical copula process (2009)
0 of 32 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2601 linked papers.