pith. sign in

arxiv: 2606.27046 · v1 · pith:AN3HHOGDnew · submitted 2026-06-25 · 📊 stat.ME · q-fin.MF

Conditional Leibniz Derivative Estimation with an Application to American Call Min-Options

Xingyu Ren , Michael C. Fu , Pierre L'Ecuyer This is my paper

Pith reviewed 2026-06-26 02:37 UTC · model grok-4.3

classification 📊 stat.ME q-fin.MF
keywords Leibniz derivative estimationconditional Monte Carlolikelihood ratio methodAmerican optionsmin-optionvariance reductiondiscontinuous performance
0
0 comments X

The pith

Recursive conditioning applied to Leibniz rules produces a derivative estimator free of likelihood ratio terms for discontinuous performance functions.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a conditional Leibniz estimator by combining recursive conditioning with the Leibniz integral representation of derivatives. This removes the likelihood ratio component that normally causes variance to grow with input dimension in Monte Carlo derivative estimation. The method targets models with discontinuous sample performance, such as American call min-options. It keeps the estimator unbiased while remaining simple to implement. Simulation results on the min-option model confirm low variance relative to standard approaches.

Core claim

By applying recursive conditioning to the Leibniz integral representation of the derivative, the resulting conditional Leibniz estimator eliminates LR terms entirely. The estimator therefore avoids the linear growth of variance with the dimension of the stochastic input that affects conventional LR-based Leibniz estimators, while retaining unbiasedness for discontinuous performance functions arising in the American call min-option model.

What carries the argument

The conditional Leibniz estimator obtained by recursive conditioning of the Leibniz integral representation, which removes all likelihood ratio factors from the derivative estimator.

If this is right

  • The estimator applies directly to other high-dimensional stochastic models whose performance functions contain discontinuities.
  • Implementation requires only the ability to condition the underlying random variables recursively, without needing explicit likelihood ratios.
  • For the American call min-option, the method delivers derivative estimates whose variance does not increase with the number of underlying assets.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The approach may extend to other path-dependent options whose exercise decisions create discontinuities at random times.
  • Because the estimator stays unbiased under the stated conditioning, it could serve as a building block for combining with control variates or other variance-reduction techniques.
  • In models where the input dimension is large, the absence of LR terms suggests computational savings that grow with problem size.

Load-bearing premise

Recursive conditioning can be applied to the Leibniz integral representation without introducing bias or additional variance sources that would invalidate the estimator for discontinuous performance functions.

What would settle it

Run the conditional Leibniz estimator on the American call min-option model alongside an unbiased reference method and observe whether its sample variance grows with input dimension or its bias exceeds Monte Carlo error.

Figures

Figures reproduced from arXiv: 2606.27046 by Michael C. Fu, Pierre L'Ecuyer, Xingyu Ren.

Figure 1
Figure 1. Figure 1: Illustration of the optimal early exercise region in a two-dimensional American call min-option [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗
read the original abstract

Leibniz derivative estimation is a Monte Carlo technique for estimating derivatives of a discontinuous sample performance in stochastic models with respect to parameters of interest. By combining the push-out likelihood ratio (LR) method with Leibniz integral rules, it generalizes a broad class of existing LR-based derivative estimators. However, as an LR-based method, its variance is often higher than that of perturbation analysis-based methods and may grow linearly with the dimension of the stochastic input whose distribution depends on the parameter. In this paper, we propose a recursive conditioning approach and combine it with the Leibniz derivative estimation framework. The resulting conditional Leibniz estimator does not involve LR terms and therefore is not subject to variance growth with the input dimension. It also has a simple form and is easy to implement. We apply the method to an American call min-option model, and simulation results show its effectiveness and low-variance performance.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper proposes combining recursive conditioning with the Leibniz integral rule within a derivative estimation framework to produce a 'conditional Leibniz estimator' for gradients of discontinuous Monte Carlo sample performances. The central claim is that the resulting estimator is free of likelihood-ratio terms (hence free of dimension-dependent variance growth), remains unbiased, has a simple closed form, and is easy to implement. The method is applied to pricing and sensitivity analysis of an American call min-option, with simulation results asserted to demonstrate effectiveness and low variance.

Significance. If the recursive conditioning step can be shown to permit unbiased interchange of derivative and expectation for the discontinuous payoff induced by the min operator and optimal exercise boundary, the approach would supply a practical, low-variance alternative to standard LR or pathwise estimators in high-dimensional discontinuous settings. The absence of any LR score function is a potentially attractive feature for dimension scaling.

major comments (3)
  1. [derivation of conditional Leibniz estimator (likely §3)] The manuscript provides no theorem or explicit derivation establishing that the chosen recursive conditioning sequence permits exact interchange of derivative and integral at each step without residual score-function terms or bias at the discontinuity surfaces created by the min operator and early-exercise decision. This interchange is load-bearing for the unbiasedness claim.
  2. [§4–5, American call min-option model] The application to the American call min-option (presumably §4–5) invokes the estimator on a payoff that is discontinuous in both the state variables and the exercise decision, yet no verification is given that the conditioning variables are chosen so that the conditional densities remain differentiable and the Leibniz boundary terms vanish pathwise.
  3. [simulation study (likely §6)] Simulation results are cited as evidence of low variance, but the text supplies neither the number of replications, the precise variance-reduction ratios versus plain LR or IPA estimators, nor the input dimension at which the comparison is performed, rendering the central empirical claim unverifiable.
minor comments (2)
  1. Notation for the recursive conditioning operator and the resulting estimator should be introduced with an explicit equation rather than prose description only.
  2. [abstract and introduction] The abstract states that the estimator 'does not involve LR terms,' but the manuscript should clarify whether this holds exactly or only after an additional approximation whose error is controlled.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We appreciate the referee's detailed review and the opportunity to clarify and strengthen the manuscript. We address each of the major comments below, agreeing that additional formalization and details are warranted for the claims of unbiasedness and empirical performance.

read point-by-point responses

  1. Referee: The manuscript provides no theorem or explicit derivation establishing that the chosen recursive conditioning sequence permits exact interchange of derivative and integral at each step without residual score-function terms or bias at the discontinuity surfaces created by the min operator and early-exercise decision. This interchange is load-bearing for the unbiasedness claim.

    Authors: We acknowledge that the current manuscript lacks a formal theorem for the interchange under recursive conditioning. In the revision, we will include a new theorem (likely in Section 3) that rigorously proves the unbiasedness of the conditional Leibniz estimator by showing that the recursive conditioning allows exact interchange without introducing LR terms or bias at discontinuities. The proof will rely on the properties of the chosen conditioning sequence and the differentiability of the conditional densities. revision: yes

  2. Referee: The application to the American call min-option (presumably §4–5) invokes the estimator on a payoff that is discontinuous in both the state variables and the exercise decision, yet no verification is given that the conditioning variables are chosen so that the conditional densities remain differentiable and the Leibniz boundary terms vanish pathwise.

    Authors: We agree that explicit verification is needed. In the revised manuscript, we will add a subsection in §4 or §5 detailing the choice of conditioning variables for the American call min-option model. This will demonstrate that the conditional densities are differentiable with respect to the parameters and that the boundary terms from the Leibniz rule vanish pathwise due to the recursive structure, ensuring no residual bias. revision: yes

  3. Referee: Simulation results are cited as evidence of low variance, but the text supplies neither the number of replications, the precise variance-reduction ratios versus plain LR or IPA estimators, nor the input dimension at which the comparison is performed, rendering the central empirical claim unverifiable.

    Authors: We will revise the simulation study in §6 to include the missing details: the number of replications used, the exact variance-reduction ratios compared to LR and IPA estimators, and the input dimensions tested. This will make the empirical claims fully verifiable and strengthen the evidence for the low-variance performance of the conditional Leibniz estimator. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation combines established LR and Leibniz techniques

full rationale

The abstract derives the conditional Leibniz estimator by combining the push-out likelihood ratio method with Leibniz integral rules and a recursive conditioning approach. This is presented as generalizing existing LR-based estimators without any indication that the central unbiasedness or variance claims reduce by construction to fitted parameters, self-definitional equations, or load-bearing self-citations. The derivation is self-contained against external benchmarks of the cited methods.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the applicability of the Leibniz integral rule to the discontinuous performance function and on the existence of a recursive conditioning decomposition that removes the LR weight without bias. No free parameters or invented entities are mentioned.

axioms (2)
  • domain assumption The Leibniz integral rule can be applied to interchange differentiation and expectation for the discontinuous sample performance function.
    Invoked to generalize existing LR-based derivative estimators.
  • ad hoc to paper A recursive conditioning decomposition exists that eliminates the likelihood ratio term while preserving the derivative estimate.
    This is the key step introduced in the paper to create the conditional estimator.

pith-pipeline@v0.9.1-grok · 5683 in / 1304 out tokens · 42977 ms · 2026-06-26T02:37:36.208049+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

16 extracted references · 1 canonical work pages

  1. [1]

    Augmenting automatic differentiation for a single-server queue via the

    Fu, Michael C , journal=. Augmenting automatic differentiation for a single-server queue via the

  2. [2]

    arXiv preprint arXiv:2511.00006 , year=

    Stochastic derivative estimation for discontinuous sample performances: A Leibniz integration perspective , author=. arXiv preprint arXiv:2511.00006 , year=

    work page arXiv

  3. [3]

    Journal of the Operational Research Society , volume=

    Optimal acceptance of incompatible kidneys , author=. Journal of the Operational Research Society , volume=. 2025b , publisher=

  4. [4]

    A unified view of the

    L'Ecuyer, Pierre , journal=. A unified view of the. 1990 , publisher=

    1990

  5. [5]

    Annals of Operations Research , volume=

    Sensitivity analysis of discrete event systems by the push out method , author=. Annals of Operations Research , volume=. 1992 , publisher=

    1992

  6. [6]

    Operations Research , volume=

    A new stochastic derivative estimator for discontinuous payoff functions with application to financial derivatives , author=. Operations Research , volume=. 2012 , publisher=

    2012

  7. [7]

    Operations Research , volume=

    A new unbiased stochastic derivative estimator for discontinuous sample performances with structural parameters , author=. Operations Research , volume=. 2018 , publisher=

    2018

  8. [8]

    Variance reduction for generalized likelihood ratio method by conditional

    Peng, Yijie and Fu, Michael C and Hu, Jiaqiao and L’Ecuyer, Pierre and Tuffin, Bruno , journal=. Variance reduction for generalized likelihood ratio method by conditional. 2022 , publisher=

    2022

  9. [9]

    European Journal of Operational Research , volume=

    Generalized likelihood ratio method for stochastic models with uniform random numbers as inputs , author=. European Journal of Operational Research , volume=. 2025 , publisher=

    2025

  10. [10]

    Conditional

    Fu, Michael C and Hu, Jian-Qiang , year=. Conditional

  11. [11]

    Sensitivity analysis for

    Fu, Michael C and Hu, Jian-Qlang , journal=. Sensitivity analysis for. 1995 , publisher=

    1995

  12. [12]

    2022 , publisher=

    L’Ecuyer, Pierre and Puchhammer, Florian and Ben Abdellah, Amal , journal=. 2022 , publisher=

    2022

  13. [13]

    The valuation of

    Detemple, Jerome and Feng, Shui and Tian, Weidong , journal=. The valuation of. 2003 , publisher=

    2003

  14. [14]

    2011 , publisher=

    The Geometry of Physics: An Introduction , author=. 2011 , publisher=

    2011

  15. [15]

    Option pricing via

    Heidergott, Bernd , journal=. Option pricing via. 2001 , publisher=

    2001

  16. [16]

    write newline

    " write newline "" before.all 'output.state := FUNCTION article output.bibitem format.authors "author" output.check author format.key output output.year.check new.block format.title "title" output.check new.block crossref missing format.jour.vol output format.article.crossref output.nonnull format.pages output if new.block note output fin.entry format.ann...