pith. sign in

arxiv: 2606.28891 · v1 · pith:V7OYE3NLnew · submitted 2026-06-27 · 💱 q-fin.MF

Hedging Maturity-Specific Risk in Forward Curve Derivatives under Stochastic Volatility

Riccardo Alberti , Sven Karbach This is my paper

Pith reviewed 2026-06-30 08:35 UTC · model grok-4.3

classification 💱 q-fin.MF
keywords variance-optimal hedgingforward curvesstochastic volatilityHJMM modelquadratic hedging errorGaltchouk-Kunita-Watanabe projectiondensity of strategiesresidual risk
0
0 comments X

The pith

The quadratic hedging error for forward-curve claims splits exactly into bucket, rank and residual components under infinite-rank stochastic volatility.

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

The paper shows that variance-optimal hedging of European claims written on forward curves is given by the Galtchouk-Kunita-Watanabe projection onto the covariance-norm quotient of the forward-curve martingale. It establishes density of finite-maturity and delivery-window strategies, convergence of spectral finite-rank approximations, and an exact three-way split of the quadratic error. In enlarged filtrations the residual term becomes a stochastic-volatility floor whenever the claim loads on non-traded covariance noise. The results are illustrated inside affine stochastic-covariance and multiplicative HJMM models, including a concrete CIR example.

Core claim

Under Heath-Jarrow-Morton-Musiela dynamics modulated by an infinite-rank stochastic covariance, the variance-optimal hedge is the Galtchouk-Kunita-Watanabe projection with respect to the covariance-norm quotient; the quadratic hedging error then decomposes exactly into bucket, rank and residual risk, and the residual risk functions as a stochastic-volatility floor in enlarged filtrations for claims that depend on non-traded covariance noise.

What carries the argument

The Galtchouk-Kunita-Watanabe projection onto the covariance-norm quotient generated by the forward-curve martingale, applied to an infinite-rank stochastic covariance component in the HJMM SPDE.

If this is right

  • Finite-maturity and delivery-window strategies are dense in the hedging space.
  • Spectral finite-rank projections converge to the optimal hedge.
  • The quadratic error decomposes exactly into bucket, rank and residual risk.
  • Residual risk acts as a stochastic-volatility floor in enlarged filtrations for claims on non-traded covariance noise.

Where Pith is reading between the lines

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

  • Traders could monitor bucket, rank and residual exposures separately when managing large forward-curve books.
  • The decomposition suggests a natural way to allocate capital across different sources of unhedgeable risk in infinite-dimensional markets.
  • Similar projections might be examined in other infinite-dimensional term-structure models that carry stochastic volatility.

Load-bearing premise

The forward curve follows a Heath-Jarrow-Morton-Musiela stochastic partial differential equation modulated by an infinite-rank stochastic covariance component that makes the Galtchouk-Kunita-Watanabe projection well-defined.

What would settle it

A concrete model in which the quadratic hedging error cannot be decomposed into the three stated components, or in which finite-maturity strategies fail to be dense, would falsify the central claims.

Figures

Figures reproduced from arXiv: 2606.28891 by Riccardo Alberti, Sven Karbach.

Figure 1
Figure 1. Figure 1: The two panels separate the two convergence questions of Proposition 3.18. Left (rank convergence): with the full continuum of maturities available, the rank-only er￾ror (ii)+(iii), rank gap plus floor, decays monotonically as covariance factors are retained and converges down to the stochastic-volatility floor (iii) E[N 2 T ]. Right (implementation convergence): at a fixed spectral rank N∗ = 8, refining a… view at source ↗
Figure 2
Figure 2. Figure 2: Covariance-weighted truncation gap P k /∈S akw 2 k under variance (principal￾component) selection versus the hedging-norm-optimal selection, for a claim that also loads on a low-variance factor ek0 (k0 = 13, dotted line). Ranking factors by explained variance leaves the hedging-relevant factor out until n ≥ k0; the hedging norm selects it immediately. The plotted quantity is purely spectral (no Monte-Carlo… view at source ↗
read the original abstract

We study the variance-optimal hedging of European contingent claims written on forwards. We assume that the dynamics of the underlying forward curves follow a Heath--Jarrow--Morton--Musiela stochastic partial differential equation modulated by an infinite-rank stochastic covariance component. The variance-optimal hedge is then given by the Galtchouk--Kunita--Watanabe projection with respect to some covariance-norm quotient generated by the forward curve martingale. We show density of finite-maturity and delivery-window strategies, convergence of spectral finite-rank hedge projections and an exact decomposition of the quadratic hedging error into bucket, rank and residual risk components. In enlarged filtrations, the residual risk is a stochastic-volatility floor for claims loading on non-traded covariance noise. We illustrate the hedging framework in affine stochastic covariance and multiplicative HJMM models, and give a concrete example of the decomposition in a CIR stochastic covariance model.

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 develops a variance-optimal hedging framework for European claims on forward curves whose dynamics are given by an HJMM-Musiela SPDE driven by an infinite-rank stochastic covariance process. The hedge is obtained via the Galtchouk-Kunita-Watanabe projection onto the covariance-norm quotient generated by the forward-curve martingale. The central results are the density of finite-maturity and delivery-window strategies, L2-convergence of spectral finite-rank projections, and an exact orthogonal decomposition of the quadratic hedging error into bucket, rank, and residual components; the residual is interpreted as a stochastic-volatility floor when the filtration is enlarged by non-traded covariance noise. The framework is illustrated in affine stochastic-covariance and multiplicative HJMM models, with a concrete CIR example.

Significance. If the operator-theoretic foundations are rigorously justified, the exact bucket/rank/residual decomposition supplies a transparent separation of hedgeable and unhedgeable risks in an infinite-dimensional stochastic-volatility setting. This is a substantive contribution to the mathematical-finance literature on curve hedging, as it moves beyond finite-factor approximations and quantifies the residual floor due to covariance noise. The model illustrations add concreteness without introducing free parameters.

major comments (2)
  1. [§2.2] §2.2, Definition 2.3 and Assumption 2.4: The existence of the Galtchouk-Kunita-Watanabe projection with respect to the covariance-norm quotient is asserted for an infinite-rank stochastic covariance operator on the Musiela space, yet no pathwise regularity (e.g., nuclear or Hilbert-Schmidt property of the covariance operator) or predictable-representation condition is supplied to guarantee that the quotient is a well-defined Hilbert space and that the projection exists. All subsequent claims on density, spectral convergence, and the exact decomposition rest on this step; without the missing operator conditions the results do not follow.
  2. [§4.1] §4.1, Theorem 4.2: The claimed L2-convergence of the spectral finite-rank hedge projections is stated without an explicit rate or error bound that would control the tail of the infinite-rank covariance; the proof sketch appears to rely on the same unverified quotient-space construction.
minor comments (2)
  1. [Eq. (3.7)] The distinction between the bucket-risk and rank-risk components in the decomposition (Eq. (3.7)) would be clearer if the orthogonal projection onto the finite-maturity subspace were written explicitly rather than left implicit.
  2. [§2] Notation for the Musiela parametrization and the covariance-norm quotient is introduced in §2 but reused with slight variations in §5; a single consolidated table of symbols would improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thorough review and valuable comments on the operator-theoretic foundations of our framework. We agree that the manuscript requires additional assumptions to rigorously establish the Galtchouk-Kunita-Watanabe projection in the infinite-dimensional setting. We will revise the paper to include the necessary regularity conditions and error bounds. Below we address each major comment.

read point-by-point responses

  1. Referee: [§2.2] §2.2, Definition 2.3 and Assumption 2.4: The existence of the Galtchouk-Kunita-Watanabe projection with respect to the covariance-norm quotient is asserted for an infinite-rank stochastic covariance operator on the Musiela space, yet no pathwise regularity (e.g., nuclear or Hilbert-Schmidt property of the covariance operator) or predictable-representation condition is supplied to guarantee that the quotient is a well-defined Hilbert space and that the projection exists. All subsequent claims on density, spectral convergence, and the exact decomposition rest on this step; without the missing operator conditions the results do not follow.

    Authors: We acknowledge this gap in the current version. To address it, we will introduce a new assumption (Assumption 2.5) requiring that the stochastic covariance operator process is predictable and takes values in the Hilbert-Schmidt operators on the Musiela space almost surely. This ensures the covariance-norm quotient is a separable Hilbert space, allowing the application of the Galtchouk-Kunita-Watanabe decomposition theorem in Hilbert spaces. With this, the projection exists uniquely, and the subsequent density results, spectral convergence, and orthogonal decomposition follow directly from the projection properties and the spectral theorem for compact operators. We will also add a remark on the predictable representation property holding under the affine model assumptions used in the illustrations. revision: yes

  2. Referee: [§4.1] §4.1, Theorem 4.2: The claimed L2-convergence of the spectral finite-rank hedge projections is stated without an explicit rate or error bound that would control the tail of the infinite-rank covariance; the proof sketch appears to rely on the same unverified quotient-space construction.

    Authors: We agree that providing an explicit convergence rate would improve the result. In the revised manuscript, we will strengthen Theorem 4.2 by deriving an error bound: the L2-distance between the infinite-rank and N-rank projections is bounded by the square root of the sum of the tail eigenvalues of the covariance operator, i.e., ||P - P_N|| <= (sum_{k>N} lambda_k)^{1/2}, where lambda_k are the eigenvalues. This bound follows from the Hilbert-Schmidt property and controls the tail under the assumption that the covariance is trace-class. The proof will be updated to explicitly use the new Assumption 2.5. revision: yes

Circularity Check

0 steps flagged

No circularity; derivation self-contained under standard stochastic assumptions

full rationale

The paper assumes HJMM-Musiela SPDE dynamics with infinite-rank stochastic covariance and posits existence of the GKW projection in the induced covariance-norm quotient; all subsequent claims (density of strategies, spectral convergence, bucket/rank/residual decomposition) are derived from this setup via standard stochastic calculus without any reduction of outputs to fitted inputs, self-definitional loops, or load-bearing self-citations. No equations or steps in the provided abstract or description equate a claimed result to its own construction by renaming or fitting.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claims rest on the modeling choice of an infinite-rank stochastic covariance in the HJMM-Musiela SPDE and on the existence of the GKW projection in the associated covariance-norm quotient space; these are domain assumptions rather than derived results.

axioms (2)
  • domain assumption Forward curve dynamics follow a Heath-Jarrow-Morton-Musiela SPDE modulated by an infinite-rank stochastic covariance component
    Explicitly stated as the modeling framework in the abstract.
  • domain assumption The Galtchouk-Kunita-Watanabe projection exists with respect to the covariance-norm quotient generated by the forward curve martingale
    Required for the variance-optimal hedge to be well-defined; invoked without further justification in the abstract.

pith-pipeline@v0.9.1-grok · 5676 in / 1445 out tokens · 32755 ms · 2026-06-30T08:35:36.604635+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

37 extracted references · 2 canonical work pages · 2 internal anchors

  1. [1]

    Andresen, S

    A. Andresen, S. Koekebakker, and S. Westgaard. Modeling electricity forward prices using the multivariate normal inverse gaussian distribution.Journal of Energy Markets, 3(3):3–25, sep 2010

    2010

  2. [2]

    Bátkai, B

    A. Bátkai, B. Farkas, P. Csomós, and A. Ostermann. Operator semigroups for numerical analysis. Lecture notes for the 15th Internet Seminar, 2011/12, 2012. Accessed: 2024-08-13

    2011

  3. [3]

    F. E. Benth and H. Eyjolfsson. Robustness of Hilbert space-valued stochastic volatility models. Finance and Stochastics, 28(4):1117–1146, 2024

    2024

  4. [4]

    F. E. Benth and S. Koekebakker. Stochastic modeling of financial electricity contracts.Energy Economics, 30(3):1116–1157, 2008

    2008

  5. [5]

    F. E. Benth and P. Krühner. Representation of infinite-dimensional forward price models in com- modity markets.Communications in Mathematics and Statistics, 2(1):47–106, 2014

    2014

  6. [6]

    F. E. Benth, B. Rüdiger, and A. Süss. Ornstein–Uhlenbeck processes in Hilbert space with non- Gaussian stochastic volatility.Stochastic Processes and their Applications, 128(2):461–486, 2018

    2018

  7. [7]

    F. E. Benth, J. Šaltyt˙ e Benth, and S. Koekebakker.Stochastic Modeling of Electricity and Related Markets. World Scientific Publishing Co. Pte. Ltd., 2008

    2008

  8. [8]

    F. E. Benth and I. C. Simonsen. The Heston stochastic volatility model in Hilbert space.Stochastic Analysis and Applications, 36(4):733–750, 2018

    2018

  9. [9]

    Björk, G

    T. Björk, G. Di Masi, Y. Kabanov, and W. Runggaldier. Towards a general theory of bond markets. Finance and Stochastics, 1(2):141–174, 1997

    1997

  10. [10]

    Borak and R

    S. Borak and R. Weron. A semiparametric factor model for electricity forward curve dynamics. Journal of Energy Markets, 1(3):3–16, 2008

    2008

  11. [11]

    R. A. Carmona.HJM: A Unified Approach to Dynamic Models for Fixed Income, Credit and Equity Markets, pages 1–50. Springer Berlin Heidelberg, Berlin, Heidelberg, 2007

    2007

  12. [12]

    R. A. Carmona and M. R. Tehranchi.Interest rate models: an infinite dimensional stochastic analysis perspective. Berlin: Springer, 2006

    2006

  13. [13]

    Chatziandreou and S

    K. Chatziandreou and S. Karbach. Semi-static variance-optimal hedging of covariance risk in multi- asset derivatives, 2026

    2026

  14. [14]

    R. Cont. Modeling term structure dynamics: an infinite dimensional approach.International Journal of Theoretical and Applied Finance, 8(3):357–380, 2005

    2005

  15. [15]

    S. Cox, S. Karbach, and A. Khedher. An infinite-dimensional affine stochastic volatility model. Mathematical Finance, 32(3):878–906, 2022

    2022

  16. [16]

    GeneralizedFellerprocessesandMarkovianliftsofstochasticVolterra processes: the affine case.J

    C.CuchieroandJ.Teichmann. GeneralizedFellerprocessesandMarkovianliftsofstochasticVolterra processes: the affine case.J. Evol. Equ., 20(4):1301–1348, 2020

    2020

  17. [17]

    Da Prato and J

    G. Da Prato and J. Zabczyk.Stochastic Equations in Infinite Dimensions, volume 152 ofEncyclo- pedia of Mathematics and its Applications. Cambridge University Press, Cambridge, second edition, 2014

    2014

  18. [18]

    De Donno, P

    M. De Donno, P. Guasoni, and M. Pratelli. Super-replication and utility maximization in large financial markets.Stochastic Processes and their Applications, 115(12):2006–2022, 2005

    2006

  19. [19]

    De Donno and M

    M. De Donno and M. Pratelli. A theory of stochastic integration for bond markets.The Annals of Applied Probability, 15(4):2773–2791, 2005

    2005

  20. [20]

    Detering and S

    N. Detering and S. Lavagnini. A class of locally state-dependent models for forward curves, 2025

    2025

  21. [21]

    R. Douady. Yield curve smoothing and residual variance of fixed income positions. In Y. Kabanov, M. Rutkowski, and T. Zariphopoulou, editors,Inspired by Finance: The Musiela Festschrift, pages 221–256. Springer International Publishing, Cham, 2014

    2014

  22. [22]

    Ekeland and E

    I. Ekeland and E. Taflin. A theory of bond portfolios.The Annals of Applied Probability, 15(2):1260– 1305, 2005

    2005

  23. [23]

    Filipović.Consistency problems for Heath-Jarrow-Morton interest rate models, volume 1760

    D. Filipović.Consistency problems for Heath-Jarrow-Morton interest rate models, volume 1760. Berlin: Springer, 2001

    2001

  24. [24]

    D. Frestad. Common and unique factors influencing daily swap returns in the nordic electricity market, 1997-2005.Energy Economics, 30(3):1081–1097, May 2008

    1997

  25. [25]

    Friesen and S

    M. Friesen and S. Karbach. Stationary covariance regime for affine stochastic covariance models in Hilbert spaces.Finance Stoch., 28(4):1077–1116, 2024

    2024

  26. [26]

    J. He, S. Karbach, and A. Khedher. Pricing Options on Forwards in Function-Valued Affine Sto- chastic Volatility Models. Preprint, arXiv:2508.14813 [q-fin.MF] (2025), 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  27. [27]

    Kallsen and A

    J. Kallsen and A. Pauwels. Variance-optimal hedging in general affine stochastic volatility models. Advances in Applied Probability, 42(1):83–105, 2010

    2010

  28. [28]

    S. Karbach. Heat modulated affine stochastic volatility models for forward curve dynamics.arXiv preprint arXiv:2409.13070, 2024. HEDGING MATURITY-SPECIFIC RISK IN FOR W ARD CUR VE DERIV ATIVES 27

    work page internal anchor Pith review arXiv 2024

  29. [29]

    S. Karbach. Finite-rank approximation of affine processes on positive Hilbert-Schmidt operators.J. Math. Anal. Appl., 553(2):35, 2026. Id/No 129852

    2026

  30. [30]

    Koekebakker and F

    S. Koekebakker and F. Ollmar. Forward curve dynamics in the nordic electricity market.Managerial Finance, 31:73–94, 2005

    2005

  31. [31]

    J. Y. Ouvrard. Représentation de martingales vectorielles de carré intégrable à valeurs dans des espaces de hilbert réels séparables.Zeitschrift für Wahrscheinlichkeitstheorie und Verwandte Gebiete, 33:195–208, 1975

    1975

  32. [32]

    Peszat and J

    S. Peszat and J. Zabczyk.Stochastic Partial Differential Equations with Lévy Noise. An Evolution Equation Approach, volume 113 ofEncyclopedia of Mathematics and its Applications. Cambridge University Press, Cambridge, 2007

    2007

  33. [33]

    H. Pham. On quadratic hedging in continuous time.Mathematical Methods of Operations Research, 51(2):315–339, 2000

    2000

  34. [34]

    Schweizer

    M. Schweizer. A guided tour through quadratic hedging approaches. In E. Jouini, J. Cvitanic, and M. Musiela, editors,Option Pricing, Interest Rates and Risk Management, Handbooks in Mathe- matical Finance, page 538–574. Cambridge University Press, 2001

    2001

  35. [35]

    E. Taflin. Bond market completeness and attainable contingent claims.Finance and Stochastics, 9(3):429–452, 2005

    2005

  36. [36]

    S. Tappe. Compact embeddings for spaces of forward rate curves.Abstr. Appl. Anal., 2013:Article ID 709505, 6 pages, 2013

    2013

  37. [37]

    Černý and J

    A. Černý and J. Kallsen. On the structure of general mean-variance hedging strategies.The Annals of Probability, 35(4):1479–1531, 2007

    2007