pith. sign in

arxiv: 2605.17142 · v1 · pith:OCUHVDPJnew · submitted 2026-05-16 · 💱 q-fin.MF · math.PR

On the Structural Foundations of Signature Volatility Models: Existence, Arbitrage, Completeness, and the Hedging-Error Decomposition

Akmal Xodarev This is my paper

Pith reviewed 2026-05-20 15:01 UTC · model grok-4.3

classification 💱 q-fin.MF math.PR
keywords signature volatilitytensor algebrastochastic exponentialmarket completenesshedging errorno free lunchrough pathsSDE existence
0
0 comments X

The pith

Signature volatility models have global solutions and arbitrage-free pricing on an admissible weighted tensor algebra that also determines market completeness.

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

The paper establishes that signature SDEs driven by Brownian motion on a weighted tensor algebra admit unique global strong solutions when a summability condition and an exponential-integrability condition hold. These same conditions make the stochastic exponential a true martingale under the reference measure, which delivers no free lunch with vanishing risk. Market completeness on the price filtration is then characterized by the density of finite-order signature spans in the space of square-integrable claims, and the minimal such order is identified as the completeness depth. For any square-integrable payoff the hedging error decomposes into an integral against the price process plus a residual term whose size is controlled by the projection onto signature components beyond that depth.

Core claim

The admissible weighted tensor algebra on which the stochastic exponential is a true martingale and finite signature transforms do not explode is the natural valuation cell of a signature SDE, simultaneously guaranteeing global existence of solutions, absence of arbitrage, completeness at a finite signature depth, and an explicit bound on hedging residuals.

What carries the argument

The admissible weighted tensor algebra T_w equipped with summability condition H1 and exponential-integrability condition H3, which together ensure the signature SDE dS_t = S_t ⟨ℓ, Ŵ_t⟩ dB_t has global strong solutions and that the reference-measure stochastic exponential is a true martingale.

If this is right

  • Global existence and uniqueness of strong solutions hold for the infinite-dimensional signature SDE under the stated conditions.
  • The reference-measure stochastic exponential is a true martingale, yielding NFLVR on the natural filtration of the prolonged signature.
  • Market completeness on the price filtration is equivalent to density of the truncated signature span up to the completeness depth in L²(ℱ^S_T, ℚ).
  • Any square-integrable payoff admits a hedging-error decomposition whose residual is the Gram projection onto signature components beyond the completeness depth.

Where Pith is reading between the lines

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

  • The same tensor-algebra conditions may supply a template for verifying well-posedness in other rough-path or path-signature models of volatility.
  • Numerical checks of the summability and integrability conditions on empirical volatility paths could serve as a practical test for whether a given signature model stays arbitrage-free.
  • The completeness depth might be computed explicitly for low-dimensional driving signals to guide truncation choices in implementation.

Load-bearing premise

The summability condition H1 together with the exponential-integrability condition H3 are enough to produce both global solutions to the signature SDE and a true martingale property for the stochastic exponential.

What would settle it

An explicit example of a driving process that satisfies H1 and H3 yet produces an exploding finite signature transform or a stochastic exponential that fails to be a true martingale would refute the sufficiency claim.

read the original abstract

We establish four structural results for signature volatility models. First, we prove global existence and uniqueness of strong solutions to the signature SDE $dS_t = S_t \langle \ell, \widehat{W}_t \rangle \, dB_t$ on the weighted tensor algebra $T_w$, identifying the admissibility class through a summability condition H1 and an exponential-integrability condition H3 for the square-integrable stochastic-exponential construction. Second, we establish the asset-pricing part on the natural filtration of the prolonged signature and separate it from transform non-explosion: H3 makes the reference-measure stochastic exponential a true martingale, hence yields NFLVR, while global solvability of the associated infinite-dimensional Riccati equation is the additional condition equivalent to absence of explosion for finite signature transforms. Third, we characterise market completeness on the price filtration via the density of the truncated signature span $\mathrm{span}\{\langle e_I, \widehat{W}_T \rangle : |I| \leq N\}$ inside $L^2(\mathcal{F}^S_T, \mathbb{Q})$, and identify the minimal such $N$, the price-filtration completeness depth. Fourth, we derive the hedging-error decomposition $X = \mathbb{E}_\mathbb{Q}[X] + \int_0^T H_s \, dS_s + \varepsilon_T$ for square-integrable payoffs, with residual expanded through the Gram projection of signature components beyond the completeness depth and bounded by a model-dependent projection error. The four results are tied by an architectural identity: the admissible weighted tensor algebra on which the stochastic exponential is a true martingale and finite signature transforms do not explode is the natural valuation cell of a signature SDE. The proofs are self-contained except for standard results from rough path theory, stochastic integration, and quadratic hedging, recalled in the appendices.

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

2 major / 2 minor

Summary. The paper establishes four structural results for signature volatility models. It proves global existence and uniqueness of strong solutions to the signature SDE dS_t = S_t ⟨ℓ, Ŵ_t⟩ dB_t on the weighted tensor algebra T_w under summability condition H1 and exponential-integrability condition H3. It separates the asset-pricing implications (H3 yields NFLVR via the reference-measure stochastic exponential being a true martingale) from non-explosion of finite signature transforms (via global solvability of the infinite-dimensional Riccati equation). It characterizes market completeness on the price filtration through density of the truncated signature span span{⟨e_I, Ŵ_T⟩ : |I| ≤ N} in L²(ℱ^S_T, ℚ) and identifies the minimal such N as the completeness depth. Finally, it derives a hedging-error decomposition X = E_ℚ[X] + ∫ H_s dS_s + ε_T for square-integrable payoffs, with the residual expressed via Gram projection of higher signature components and bounded by a model-dependent projection error. These results are unified by the claim that the admissible weighted tensor algebra is the natural valuation cell of the signature SDE. Proofs are self-contained except for recalled standard results from rough path theory, stochastic integration, and quadratic hedging.

Significance. If the central claims hold, the work supplies a rigorous infinite-dimensional foundation for signature volatility models, clarifying admissibility, no-arbitrage, completeness, and hedging-error bounds in a unified framework. This could strengthen the mathematical underpinnings of rough-path and signature methods in quantitative finance, particularly for model selection and risk management in high-dimensional or path-dependent settings.

major comments (2)
  1. [§2] §2 (Existence and global solvability): The claim that H1 (summability) together with H3 (exponential integrability) suffice for global strong solutions on the full weighted tensor algebra T_w without explosion relies on a priori estimates that close uniformly across all tensor levels. Standard Gronwall or stopping-time arguments for infinite-dimensional linear SDEs may fail to produce a uniform-in-time bound on the weighted norm unless the linear operator induced by ⟨ℓ, Ŵ_t⟩ has level-by-level norm growth controlled by H1 independently of the driving signature; if this control is only implicit or relies on comparison with the finite-dimensional Riccati, the equivalence between the admissible set and the valuation cell is not yet secured.
  2. [§3] §3 (Asset-pricing part): The separation of H3 (true-martingale property yielding NFLVR) from the additional Riccati solvability condition for non-explosion of finite transforms is stated clearly, but the manuscript does not explicitly verify that the same pair H1+H3 automatically implies the Riccati condition on the infinite-dimensional level; a concrete counter-example or additional estimate showing the implication would strengthen the architectural identity.
minor comments (2)
  1. [Introduction / §2] Notation for the weighted tensor algebra T_w and the prolonged signature Ŵ_t should be introduced with an explicit definition of the weight sequence and the norm in the first section where they appear.
  2. [§4] The hedging-error decomposition in §4 expands the residual via Gram projection; a short remark on how the projection error bound depends on the choice of basis for the signature components would improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments on our manuscript. The points raised concern the uniformity of a priori estimates in the existence proof and the explicit verification of implications between our hypotheses. We respond to each major comment below and have revised the manuscript accordingly to provide additional details.

read point-by-point responses

  1. Referee: [§2] §2 (Existence and global solvability): The claim that H1 (summability) together with H3 (exponential integrability) suffice for global strong solutions on the full weighted tensor algebra T_w without explosion relies on a priori estimates that close uniformly across all tensor levels. Standard Gronwall or stopping-time arguments for infinite-dimensional linear SDEs may fail to produce a uniform-in-time bound on the weighted norm unless the linear operator induced by ⟨ℓ, Ŵ_t⟩ has level-by-level norm growth controlled by H1 independently of the driving signature; if this control is only implicit or relies on comparison with the finite-dimensional Riccati, the equivalence between the admissible set and the valuation cell is not yet secured.

    Authors: We thank the referee for this observation, which correctly identifies the need for uniform control. The proof of Theorem 2.1 derives the a priori estimate on the weighted norm directly from H1 by bounding the level-by-level operator norms of ⟨ℓ, Ŵ_t⟩ via the summability series, which is independent of the driving path. This bound is then combined with H3 to close a Gronwall inequality uniformly in time and across all tensor degrees, without relying on finite-dimensional Riccati comparisons as the primary tool. We have added an explicit lemma (new Lemma 2.4) stating the uniform bound and clarifying that the admissible set under H1+H3 coincides with the valuation cell on which global solutions exist. This makes the argument self-contained at the infinite-dimensional level. revision: yes

  2. Referee: [§3] §3 (Asset-pricing part): The separation of H3 (true-martingale property yielding NFLVR) from the additional Riccati solvability condition for non-explosion of finite transforms is stated clearly, but the manuscript does not explicitly verify that the same pair H1+H3 automatically implies the Riccati condition on the infinite-dimensional level; a concrete counter-example or additional estimate showing the implication would strengthen the architectural identity.

    Authors: We agree that an explicit link strengthens the presentation. While the separation of H3 (for the martingale property and NFLVR) from Riccati solvability (for non-explosion of finite transforms) is deliberate, the global solvability of the infinite-dimensional Riccati equation follows from the same a priori weighted-norm estimates established under H1+H3 in Section 2. We have added a short paragraph and a remark in Section 3 that derives the Riccati bound by direct comparison with the evolution of the weighted norm, showing that H1 controls the quadratic terms levelwise and H3 prevents explosion. No counter-example arises under these hypotheses, as the estimates are uniform. revision: yes

Circularity Check

0 steps flagged

No circularity; derivation rests on external standard results

full rationale

The paper states that its four structural results (global existence via H1/H3, NFLVR from martingale property, completeness via truncated signature density, and hedging-error decomposition) are proved self-containedly except for standard results from rough path theory, stochastic integration, and quadratic hedging recalled in appendices. The architectural identity equating the admissible weighted tensor algebra to the natural valuation cell is derived from the summability and exponential-integrability conditions plus global solvability of the infinite-dimensional Riccati equation, without any quoted reduction of a prediction or central claim to a fitted input, self-definition, or unverified self-citation chain. No load-bearing step reduces by construction to its own inputs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper relies on standard results from rough path theory, stochastic integration, and quadratic hedging that are recalled in the appendices. The main additions are the identification of admissibility conditions H1 and H3 and the architectural identity linking the four results.

axioms (1)
  • standard math Standard results from rough path theory, stochastic integration, and quadratic hedging
    Recalled in the appendices as stated in the abstract.

pith-pipeline@v0.9.0 · 5879 in / 1392 out tokens · 74780 ms · 2026-05-20T15:01:37.883005+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

48 extracted references · 48 canonical work pages · 1 internal anchor

  1. [1]

    \ Gérard, L A

    Eduardo Abi Jaber and Louis-Amand G\'erard. Signature volatility models: pricing and hedging with F ourier. SIAM Journal on Financial Mathematics, 16 0 (2): 0 606--642, 2025. doi:10.1137/24M1636952

    work page doi:10.1137/24m1636952 2025

  2. [2]

    Affine V olterra processes

    Eduardo Abi Jaber, Martin Larsson, and Sergio Pulido. Affine V olterra processes. Annals of Applied Probability, 29 0 (5): 0 3155--3200, 2019

    work page 2019

  3. [3]

    The quintic O rnstein-- U hlenbeck volatility model that jointly calibrates SPX and VIX smiles

    Eduardo Abi Jaber, Camille Illand, and Shaun (Xiaoyuan) Li. The quintic O rnstein-- U hlenbeck volatility model that jointly calibrates SPX and VIX smiles. arXiv preprint arXiv:2212.10917, 2023

    work page arXiv 2023

  4. [4]

    , Gérard, L A

    Eduardo Abi Jaber, Louis-Amand G\'erard, and Yuxing Huang. Path-dependent processes from signatures. arXiv preprint arXiv:2407.04956, 2024

    work page arXiv 2024

  5. [5]

    Martingale property and moment explosions in signature volatility models

    Eduardo Abi Jaber, Paul Gassiat, and Dimitri Sotnikov. Martingale property and moment explosions in signature volatility models. Finance and Stochastics, 2025. to appear; arXiv preprint arXiv:2503.17103

    work page arXiv 2025

  6. [6]

    Short-time near-the-money skew in rough fractional volatility models

    Christian Bayer, Peter Friz, Archil Gulisashvili, Blanka Horvath, and Benjamin Stemper. Short-time near-the-money skew in rough fractional volatility models. Quantitative Finance, 19 0 (5): 0 779--798, 2019

    work page 2019

  7. [7]

    The signature of a rough path: uniqueness

    Horatio Boedihardjo, Xi Geng, Terry Lyons, and Danyu Yang. The signature of a rough path: uniqueness. Advances in Mathematics, 293: 0 720--737, 2016

    work page 2016

  8. [8]

    Affine V olterra processes with jumps

    Alessandro Bondi, Giulia Livieri, and Sergio Pulido. Affine V olterra processes with jumps. Stochastic Processes and their Applications, 168: 0 104264, 2024. doi:10.1016/j.spa.2023.104264

    work page doi:10.1016/j.spa.2023.104264 2024

  9. [9]

    Deep signature transforms

    Patric Bonnier, Patrick Kidger, Imanol Perez Arribas, Cristopher Salvi, and Terry Lyons. Deep signature transforms. In Advances in Neural Information Processing Systems, 2019

    work page 2019

  10. [10]

    Iterated path integrals

    Kuo-Tsai Chen. Iterated path integrals. Bulletin of the American Mathematical Society, 83 0 (5): 0 831--879, 1977

    work page 1977

  11. [11]

    Statistical Inference for Rough Volat ility: Min- imax Theory

    Carsten H. Chong, Marc Hoffmann, Yanghui Liu, Mathieu Rosenbaum, and Gr\'egoire Szymansky. Statistical inference for rough volatility: minimax theory. Annals of Statistics, 52 0 (4): 0 1277--1306, 2024. doi:10.1214/23-AOS2343

    work page doi:10.1214/23-aos2343 2024

  12. [12]

    Stochastic analysis, rough path analysis and fractional B rownian motions

    Laure Coutin and Zhongmin Qian. Stochastic analysis, rough path analysis and fractional B rownian motions. Probability Theory and Related Fields, 122: 0 108--140, 2002

    work page 2002

  13. [13]

    Signature methods in stochastic portfolio theory

    Christa Cuchiero and Janka M\"oller. Signature methods in stochastic portfolio theory. SIAM Journal on Financial Mathematics, 16 0 (4): 0 1239--1303, 2025. doi:10.1137/24M1700223

    work page doi:10.1137/24m1700223 2025

  14. [14]

    Polynomial processes and their applications to mathematical finance

    Christa Cuchiero, Martin Keller-Ressel, and Josef Teichmann. Polynomial processes and their applications to mathematical finance. Finance and Stochastics, 16 0 (4): 0 711--740, 2012. doi:10.1007/s00780-012-0188-x

    work page doi:10.1007/s00780-012-0188-x 2012

  15. [15]

    Polynomial jump-diffusions on the unit simplex

    Christa Cuchiero, Martin Larsson, and Sara Svaluto-Ferro. Polynomial jump-diffusions on the unit simplex. Annals of Applied Probability, 31 0 (5): 0 2451--2500, 2021

    work page 2021

  16. [16]

    Signature SDEs from an affine and polynomial perspective

    Christa Cuchiero, Sara Svaluto-Ferro, and Josef Teichmann. Signature SDE s from an affine and polynomial perspective. arXiv preprint arXiv:2302.01362, 2023

    work page arXiv 2023

  17. [17]

    , Gazzani, G

    Christa Cuchiero, Guido Gazzani, Janka M\"oller, and Sara Svaluto-Ferro. Joint calibration to SPX and VIX options with signature-based models. Mathematical Finance, 35 0 (1): 0 161--213, 2025 a . doi:10.1111/mafi.12442

    work page doi:10.1111/mafi.12442 2025

  18. [18]

    , Primavera, F

    Christa Cuchiero, Francesca Primavera, and Sara Svaluto-Ferro. Universal approximation theorems for continuous functions of c\`adl\`ag paths and L \'evy-type signature models. Finance and Stochastics, 29 0 (2): 0 289--342, 2025 b . doi:10.1007/s00780-025-00557-5

    work page doi:10.1007/s00780-025-00557-5 2025

  19. [19]

    Mark H. A. Davis and Jan Ob \'oj. Market completion using options. Banach Center Publications, 83 0 (1): 0 49--60, 2008. doi:10.4064/bc83-0-4. arXiv:0710.2792

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.4064/bc83-0-4 2008

  20. [20]

    A general version of the fundamental theorem of asset pricing

    Freddy Delbaen and Walter Schachermayer. A general version of the fundamental theorem of asset pricing. Mathematische Annalen, 300: 0 463--520, 1994

    work page 1994

  21. [21]

    Une topologie sur l'espace des semimartingales

    Michel Emery. Une topologie sur l'espace des semimartingales. S\'eminaire de Probabilit\'es XIII, pages 260--280, 1979

    work page 1979

  22. [22]

    Polynomial diffusions and applications in finance

    Damir Filipovi\'c and Martin Larsson. Polynomial diffusions and applications in finance. Finance and Stochastics, 20: 0 931--972, 2016

    work page 2016

  23. [23]

    Damir Filipovi\'c, Martin Larsson, and Anders B. Trolle. Linear-rational term structure models. Journal of Finance, 72 0 (2): 0 655--704, 2017

    work page 2017

  24. [24]

    Hedging of contingent claims under incomplete information

    Hans F\"ollmer and Martin Schweizer. Hedging of contingent claims under incomplete information. In Mark H. A. Davis and Robert J. Elliott, editors, Applied Stochastic Analysis, pages 389--414. Gordon and Breach, London and New York, 1991

    work page 1991

  25. [25]

    Hedging of non-redundant contingent claims

    Hans F\"ollmer and Dieter Sondermann. Hedging of non-redundant contingent claims. In Contributions to Mathematical Economics, pages 205--223. North-Holland, 1986

    work page 1986

  26. [26]

    Asymptotics for rough stochastic volatility models

    Martin Forde and Hongzhong Zhang. Asymptotics for rough stochastic volatility models. SIAM Journal on Financial Mathematics, 8 0 (1): 0 114--145, 2017

    work page 2017

  27. [27]

    Friz and Martin Hairer

    Peter K. Friz and Martin Hairer. A Course on Rough Paths. Springer, 2014

    work page 2014

  28. [28]

    Friz and Martin Hairer

    Peter K. Friz and Martin Hairer. A Course on Rough Paths. Universitext. Springer, 2 edition, 2020

    work page 2020

  29. [29]

    Friz and Nicolas B

    Peter K. Friz and Nicolas B. Victoir. Multidimensional Stochastic Processes as Rough Paths, volume 120 of Cambridge Studies in Advanced Mathematics. Cambridge University Press, 2010

    work page 2010

  30. [30]

    Volatility is (mostly) path-dependent

    Julien Guyon and Jordan Lekeufack. Volatility is (mostly) path-dependent. Quantitative Finance, 23 0 (9): 0 1221--1258, 2023

    work page 2023

  31. [31]

    Martingales versus PDE s in finance

    David Heath and Martin Schweizer. Martingales versus PDE s in finance. Journal of Applied Probability, 37: 0 947--957, 2000

    work page 2000

  32. [32]

    Steven L. Heston. A closed-form solution for options with stochastic volatility with applications to bond and currency options. Review of Financial Studies, 6 0 (2): 0 327--343, 1993

    work page 1993

  33. [33]

    Shiryaev

    Jean Jacod and Albert N. Shiryaev. Limit Theorems for Stochastic Processes. Springer, 2 edition, 2003

    work page 2003

  34. [34]

    Large and moderate deviations for the rough B ergomi model

    Antoine Jacquier and Antoine Pannier. Large and moderate deviations for the rough B ergomi model. Stochastic Processes and their Applications, 149: 0 134--174, 2022

    work page 2022

  35. [35]

    Optimal execution with rough path signatures

    Jasdeep Kalsi, Terry Lyons, and Imanol Perez Arribas. Optimal execution with rough path signatures. SIAM Journal on Financial Mathematics, 11 0 (2): 0 470--493, 2020

    work page 2020

  36. [36]

    Ioannis Karatzas and Steven E. Shreve. Methods of Mathematical Finance. Springer, 1998

    work page 1998

  37. [37]

    David M. Kreps. Arbitrage and equilibrium in economies with infinitely many commodities. Journal of Mathematical Economics, 8 0 (1): 0 15--35, 1981

    work page 1981

  38. [38]

    On square integrable martingales

    Hiroshi Kunita and Shinzo Watanabe. On square integrable martingales. Nagoya Mathematical Journal, 30: 0 209--245, 1967

    work page 1967

  39. [39]

    System Control and Rough Paths

    Terry Lyons and Zhongmin Qian. System Control and Rough Paths. Oxford University Press, 2002

    work page 2002

  40. [40]

    Espaces de semimartingales et changement de probabilit\'e

    Jean M\'emin. Espaces de semimartingales et changement de probabilit\'e. Zeitschrift f\"ur Wahrscheinlichkeitstheorie und Verwandte Gebiete, 52: 0 9--39, 1980

    work page 1980

  41. [41]

    Sig- SDE s model for quantitative finance

    Imanol Perez Arribas, Cristopher Salvi, and Lukasz Szpruch. Sig- SDE s model for quantitative finance. In Proceedings of the First ACM International Conference on AI in Finance, 2020

    work page 2020

  42. [42]

    Philip E. Protter. Stochastic Integration and Differential Equations. Springer, 2 edition, 2005

    work page 2005

  43. [43]

    Free Lie Algebras

    Christophe Reutenauer. Free Lie Algebras. Oxford University Press, 1993

    work page 1993

  44. [44]

    Continuous Martingales and Brownian Motion

    Daniel Revuz and Marc Yor. Continuous Martingales and Brownian Motion. Springer, 3 edition, 1999

    work page 1999

  45. [45]

    The signature kernel is the solution of a G oursat PDE

    Cristopher Salvi, Thomas Cass, James Foster, Terry Lyons, and Weixin Yang. The signature kernel is the solution of a G oursat PDE . SIAM Journal on Mathematics of Data Science, 3 0 (3): 0 873--899, 2021

    work page 2021

  46. [46]

    A guided tour through quadratic hedging approaches

    Martin Schweizer. A guided tour through quadratic hedging approaches. In Option Pricing, Interest Rates and Risk Management, pages 538--574. Cambridge University Press, 2001

    work page 2001

  47. [47]

    Arbitrage et lois de martingale

    Christophe Stricker. Arbitrage et lois de martingale. Annales de l'Institut Henri Poincar\'e, 26 0 (3): 0 451--460, 1990

    work page 1990

  48. [48]

    Caract\'erisation d'une classe d'ensembles convexes de l^1 ou h^1

    Jia-An Yan. Caract\'erisation d'une classe d'ensembles convexes de l^1 ou h^1 . In S\'eminaire de Probabilit\'es XIV, pages 220--222. Springer, 1980

    work page 1980