Paper: arXiv 2502.14766

Authors: Kristoffer Andersson, Alessandro Gnoatto

Abstract

We propose a structural default model for portfolio-wide valuation adjustments (xVAs) and represent it as a system of coupled backward stochastic differential equations. The framework is divided into four layers, each capturing a key component: (i) clean values, (ii) initial margin and Collateral Valuation Adjustment (ColVA), (iii) Credit/Debit Valuation Adjustments (CVA/DVA) together with Margin Valuation Adjustment (MVA), and (iv) Funding Valuation Adjustment (FVA). Because these layers depend on one another through collateral and default effects, a naive Monte Carlo approach would require deeply nested simulations, making the problem computationally intractable. To address this challenge, we use an iterative deep BSDE approach, handling each layer sequentially so that earlier outputs serve as inputs to the subsequent layers. Initial margin is computed via deep quantile regression to reflect margin requirements over the Margin Period of Risk. We also adopt a change-of-measure method that highlights rare but significant defaults of the bank or counterparty, ensuring that these events are accurately captured in the training process. We further extend Han and Long’s (2020) a posteriori error analysis to BSDEs on bounded domains. Due to the random exit from the domain, we obtain an order of convergence of \(\mathcal{“O”}(h^{“1/4-ε”})\) rather than the usual \(\mathcal{“O”}(h^{“1/2”})\). Numerical experiments illustrate that this method drastically reduces computational demands and successfully scales to high-dimensional, non-symmetric portfolios. The results confirm its effectiveness and accuracy, offering a practical alternative to nested Monte Carlo simulations in multi-counterparty xVA analyses.

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: The paper presents advanced mathematics including deep BSDEs, measure changes, and convergence analysis with explicit O(h^{“1/4-ε”}) rates, indicating high complexity. It also demonstrates practical numerical experiments on high-dimensional portfolios, showing implementation-heavy empirical work aimed at solving real-world xVA problems.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Multi-Layer xVA<br>Structural Model"] --> B["Methodology:<br>Deep BSDE System<br>4-Layer Architecture"]
  B --> C["Data/Inputs:<br>Portfolio &<br>Market Parameters"]
  C --> D["Process:<br>Iterative Deep BSDE<br>+ Change of Measure"]
  D --> E["Process:<br>Deep Quantile<br>Regression for MVA"]
  E --> F["Process:<br>A Posteriori<br>Error Analysis"]
  F --> G["Outcomes:<br>Scalable High-Dim<br>Accurate Convergence<br>Alternative to Nested MC"]

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 structural default model for portfolio-wide valuation adjustments (xVAs) represented as a system of coupled backward stochastic differential equations.

Source sentence
We propose a structural default model for portfolio-wide valuation adjustments (xVAs) and represent it as a system of coupled backward stochastic differential equations.

Assumes 5

  • Interest rate r is either deterministic and continuous in time or described by a diffusion type SDE. model
    Source sentence
    Assumption 3.1 (Interest rate assumption). An interest rate r (including but not limited to the risk‑free rate) is either: 1. Deterministic and continuous in time; 2. Described by a diffusion type SDE.
  • Drift b and diffusion σ of the state process satisfy regularity conditions (e.g., Lipschitz, growth) as required by Assumption 3.3. model
    Source sentence
    Assumption 3.3 (Conditions on the state process). Let x_0 ∈
  • Risk measure ϱ is monotone and translation‑invariant. statistical
    Source sentence
    Assumption 4.2 (Risk measure). We assume that the risk measure ϱ is monotone and translation‑invariant.
  • Initial margin is defined as a risk measure (VaR or ES) of adverse price movements over the Margin Period of Risk (MPR). model
    Source sentence
    Initial margin is defined as a risk measure, for instance value at risk (VaR) or expected shortfall (ES), of adverse price movements of the exposure (clean value of the portfolio) over the MPR.
Show 1 more
  • The structural default model triggers default when asset values hit a pre‑defined threshold. model
    Source sentence
    A default is triggered when the value of the assets of either the bank, or the counterparty reaches a pre‑defined threshold.

Data

Synthetic data only.

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • The method achieves an order of convergence O(h^{1/4‑ε}) for BSDEs on bounded domains, slower than the usual O(h^{1/2}) due to random exit. theoretical
    Source sentence
    Due to the random exit from the domain, we obtain an order of convergence of
  • Numerical experiments show the approach drastically reduces computational demands and scales to high‑dimensional, non‑symmetric portfolios. performance
    Source sentence
    Numerical experiments illustrate that this method drastically reduces computational demands and successfully scales to high‑dimensional, non‑symmetric portfolios.
  • Empirical results confirm high accuracy by comparing to closed‑form references and nested Monte Carlo benchmarks across all four layers. performance
    Source sentence
    This is evidenced by comparisons to closed‑form references (where available) and carefully constructed nested Monte Carlo benchmarks.
  • The first error analysis for deep BSDE methods with stopping times shows a reduced O(h^{1/4‑ε}) convergence rate under suitable assumptions. theoretical
    Source sentence
    We establish the first error analysis for deep BSDE methods with stopping times, showing that domain‑restricted BSDEs retain a reduced
  • The measure‑change technique improves training stability and reduces variance by making rare default events more frequent in simulated data. theoretical
    Source sentence
    By simulating the tilted SDE X^q, one observes default scenarios more frequently. This improves training stability and reduces the variance of neural network based methods.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 54 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 1 paper in the archive

Code

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.