Pith. sign in

REVIEW 3 major objections 6 minor 2 cited by

On deep calibration of (rough) stochastic volatility models

T0 review · 3 major / 6 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper establishes a two-step pipeline in which a small neural network learns the implied volatility map of a rough volatility model and a standard optimizer then calibrates it; the authors report full-surface rough Bergomi evaluations…

desk verdict A practical, well-written two-step deep calibration paper for rough Bergomi whose main soft spot is that the synthetic validation shares the same Monte Carlo labels used in training—send it to review with a request for an independent pricing check. read the letter →

arxiv 1908.08806 v1 pith:CHON4WMJ submitted 2019-08-22 q-fin.MF

Christian Bayer , Blanka Horvath , Aitor Muguruza , Benjamin Stemper , Mehdi Tomas This is my paper · ORCID
classification q-fin.MF MSC 60G1560G2291G2091G6091B25
keywords roughvolatilityBergomideepcalibrationneuralnetworksimpliedLevenberg-MarquardtBayesianMonteCarlopricing
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper argues that the real bottleneck in calibrating rough stochastic volatility models is not the optimization step but the cost of evaluating the pricing map, and that this bottleneck can be removed by a two-step deep calibration routine. In the first step, a small fully connected neural network learns the map from model parameters to the implied volatility surface, trained on Monte Carlo prices. In the second step, a standard Levenberg-Marquardt optimizer calibrates this network approximation to market data. The authors report full-surface evaluations in about 14 microseconds, 21,000 to 35,000 times faster than Monte Carlo, and complete rough Bergomi calibration in under 40 milliseconds, with a 99% quantile RMSE below 1% across the test set and a maximum surface RMSE below 4%. The same fast pricing map also makes Bayesian parameter inference over the calibrated model computationally tractable.

What carries the argument

The load-bearing object is the neural-network approximation of the pricing map, learned in the grid-based implicit mode: the network input is the model parameter vector and the output is the full implied volatility surface on a fixed 11-by-8 grid of strikes and maturities, with spline interpolation between grid points. This architecture moves interpolation between model parameters into the network while leaving interpolation along the volatility surface to smooth splines, reducing the input dimension and the variance of the training data. The second half of the mechanism is automatic differentiation of the trained network, which supplies fast and accurate Jacobians for the Levenberg-Marquardt normal equations, and the training labels come from an efficient Monte Carlo scheme for the rough Bergomi model.

What would settle it

Take a hold-out set of rough Bergomi parameters, generate implied volatility surfaces with the same training algorithm using 60,000 paths, and compare them with surfaces from an independent high-accuracy reference such as a much larger Monte Carlo run with a different scheme; if the neural network tracks the training generator but the generator deviates from the reference by more than the reported sub-1% RMSE at short maturities or deep out-of-the-money strikes, the claim that calibration is accurate to true model prices fails.

Watch

Extended reading notes

Core claim

The central claim is that a neural network trained off-line to reproduce a model's implied volatility map can stand in for the model's slow numerical pricing engine during calibration, without sacrificing the interpretability or risk-management structure of the underlying model. The demonstration is on the rough Bergomi model, a non-Markovian stochastic volatility model whose volatility is driven by fractional Brownian motion with Hurst parameter below 1/2, making each Monte Carlo price expensive. With a three-hidden-layer, 30-neuron network whose output is an 8-by-11 grid of implied volatilities, calibrating the full rough Bergomi surface is reported to take less than 40 milliseconds; across the test set, the 99% quantile of the root-mean-square surface error is below 1% and the maximum surface RMSE is below 4%. The paper also shows that the same network enables Bayesian calibration against both synthetic and SPX market implied volatility surfaces, producing posterior distributions whose peaks lie close to the true or previously reported parameter values.

Load-bearing premise

The pipeline inherits the accuracy of the Monte Carlo scheme that produced the training labels: if that scheme is biased for rough Bergomi prices, especially at short maturities and extreme strikes where the paper reports relative errors up to 25%, the neural network learns that bias and successful synthetic calibration only shows self-consistency with the Monte Carlo generator.

Editorial extensions

If this is right

  • Rough Bergomi, which is notoriously slow to calibrate by Monte Carlo, can be calibrated in under 40 milliseconds on a standard CPU, making on-the-fly calibration practically feasible.
  • Because the network is trained on synthetic model data rather than market data, it does not need to be retrained when market regimes change; only the second optimization step is market-dependent.
  • The same two-step architecture transfers to other stochastic volatility models, with simpler networks sufficient for models such as SABR and Heston and deeper networks needed for rough models.
  • The nearly instantaneous pricing map makes Bayesian calibration practical, allowing posterior distributions over model parameters to be sampled by MCMC at very low computational cost.
  • Risk management and model interpretation remain intact because the neural network only replaces the numerical pricing engine; its outputs are still model implied volatilities with standard Jacobians.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • We infer that the same two-step recipe extends naturally to a non-constant forward variance curve, whose piecewise-constant parameters would simply enlarge the network's input dimension; the paper lists this as future work.
  • We infer that focusing training samples or loss weights on the error zones the paper reports, namely short maturities and deep out-of-the-money or in-the-money strikes, would likely reduce the maximum 25% relative errors, which occur precisely where the Monte Carlo labels are least reliable.
  • We infer that the reported speedup changes calibration from a batch end-of-day computation into an intraday or streaming task, since a 40-millisecond full-surface fit can be repeated thousands of times within a trading session.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper proposes a two-step deep calibration pipeline for stochastic volatility models, with the rough Bergomi model as the main test case. In the first step, a neural network is trained offline to approximate the model's implied-volatility map, either pointwise in option parameters or in a grid-based 'implicit' mode. In the second step, the fast surrogate is used inside a standard Levenberg-Marquardt calibration or a Bayesian MCMC procedure. The authors report a full-surface neural-network evaluation time of about 14 microseconds (21,000–35,000 times faster than their Monte Carlo benchmark), calibration times below 40 milliseconds for the rough Bergomi model, small RMSE on synthetic test surfaces, and posterior concentration near true parameters in a Bayesian experiment. They also compare this two-step approach with a one-step inverse-map neural network and find that the inverse map generalizes poorly to out-of-sample data.

Significance. If the accuracy claims hold, this is a practically valuable contribution: it separates the neural component from the calibration step, allows automatic differentiation of the pricing map, uses a small CPU-friendly network (5,668 parameters), and makes Bayesian calibration feasible for rough volatility models. The comparison of pointwise vs. grid-based training and the out-of-sample inverse-map experiment are useful for practitioners. The main caveat is that the numerical validation is currently self-referential: both the training labels and the synthetic calibration targets are produced by the same Monte Carlo scheme, so the reported accuracy is conditional on that scheme being unbiased. The paper does not provide an independent pricing benchmark or Monte Carlo error quantification, which is the key weakness for the central 'sufficient accuracy for practical use' claim.

major comments (3)
  1. [§5.1–§5.3, Figs. 2, 5, 6] The synthetic validation is circular with respect to the Monte Carlo generator. Section 5.1 states that training labels are computed with Algorithm 3.5 of Horvath, Jacquier and Muguruza [34] using 60,000 paths, and Sections 5.2 and 5.3 generate the test IV surfaces 'using Monte Carlo simulation as in Section 5.2 above'. Thus Figures 4–6 and the reported 99% RMSE quantile below 1% measure the network's ability to invert this particular Monte Carlo scheme, not its accuracy against the true rough Bergomi pricing map. The manuscript acknowledges in Section 5.1 that maximum network-vs-MC relative errors reach 25% and that these are 'consistent with the errors of the Monte Carlo training set', which explicitly makes the MC label error the floor of the reported accuracy. Since no MC bias quantification or independent pricing benchmark is provided, the central practical-accuracy claim is not yet established. Please add MC confidence intervals or a comparison with a second pricing method (e.g., the hybrid scheme of Bennedsen–Lunde–Pakkanen or asymptotic expansions) at the short-maturity and extreme-strike locations where the largest errors occur.
  2. [§5.3, Fig. 7] The SPX market-data Bayesian experiment is the only test not generated by the same Monte Carlo scheme, but it is reported only as posterior histograms. No comparison is made with parameters obtained by direct Monte Carlo calibration, no surface fit RMSE is reported, and no quantitative measure of how well the posterior matches the market data is given. As a result, Figure 7 cannot independently support the accuracy claim; it only shows that the procedure produces parameter regions that look plausible. Please report the calibrated surface error and, if possible, compare the posterior mode/median with a benchmark calibration obtained by a standard numerical pricer.
  3. [§4.2.1, Figs. 6–7] The Bayesian credible intervals depend on the assumed error scale, but the paper does not report the prior distributions or the specific values of σ_i used in the likelihood beyond 'a fractional of the spread'. The posterior widths in Figures 6–7 are therefore hard to interpret. Please state the priors and the exact heteroskedastic error specification, and include a sensitivity check to the choice of σ_i.
minor comments (6)
  1. [§5.1] The text refers to 'Algorithm 3.5 in Horvath, Jacquier and Muguruza [34]' as if it were available in this paper; since Algorithm 3.5 is not defined here, the reference should be made explicit in the sentence.
  2. [§2, Eq. (2) and Algorithm 1] The normal equations use J(µ_k) where the iteration variable µ_k is undefined; the Jacobian should presumably be evaluated at θ_k, so the notation should be J(θ_k).
  3. [§3.2.1, Eq. (5)] The network is written as F(w;θ,T,k) in equation (5) but as F(w;θ,ζ) in the surrounding text; unify the notation for readability.
  4. [§3.2.1] The sentence 'to obtain even higher accuracy, one could also choose a coarser grid, which would require longer learning time' appears to state the opposite of what is intended; likely 'finer grid' was meant.
  5. [Abstract, §5.2, references] The manuscript contains several typos: 'on the y' should be 'on the fly', 'Tabe 1' should be 'Table 1', 'accuarcy' should be 'accuracy', and in the reference list [15] 'neworks' should be 'networks' and [41] should be 'Kingma and Ba' rather than 'Kingman and Ba'.
  6. [§5.2] The statement that calibration times 'usually under 10 milliseconds' for Markovian stochastic volatility models is not supported by any experiment in this paper; please either provide data or soften the claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the two-step calibration benchmark is a held-out surrogate test, and the SPX market calibration is independent.

full rationale

The paper's derivation chain is not circular. The neural network is trained on Monte Carlo labels produced by Algorithm 3.5 of Horvath, Jacquier and Muguruza [34], and the synthetic calibration and Bayesian experiments generate test data with the same scheme, but the network is not fitted to those test parameter points: the test set is held out and the synthetic experiments measure whether the learned map, at new parameters, inverts the generator. This is a standard supervised-learning benchmark, not a fitted input renamed as a prediction. The true-model accuracy of the labels is an external numerical-analysis question, and the paper explicitly acknowledges that the maximum network-vs-MC errors are "consistent with the errors of the Monte Carlo training set," i.e., it does not claim to validate the MC scheme itself. Moreover, the SPX market-data calibration in Section 5.2/5.3 and the comparative inverse-map experiment in Appendix A provide independent, non-self-referential evidence: the market surface is not generated by the same Monte Carlo code, and the inverse-map generalization failure is measured against brute-force Monte Carlo. The self-citations to predecessor papers [7] and [35] and to the paper's own MC algorithm [34] are normal methodological references; no uniqueness theorem is imported, no ansatz is smuggled in via self-citation, and no central claim reduces by construction to its inputs. The reader's concern about MC bias is a legitimate correctness/validation risk, but it is not circularity under the criteria of this pass.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

No new physical or stochastic entities are introduced; the 'implicit/grid-based learning' is a training architecture, not an invented entity. The free parameters are the network weights, the hand-chosen training ranges, the MC path count and grid, and the architecture hyperparameters. The key domain assumptions are the accuracy of the MC generator, smoothness of the IV surface, coverage of training ranges, and the Bayesian normality assumption.

free parameters (4)
  • Neural network weights (5,668 parameters) = Trained on 34,000 Monte Carlo surfaces
    The pricing map approximation is entirely determined by these fitted weights; the central performance claims depend on this fit.
  • Training parameter ranges = xi0 in [0.01,0.16], nu in [0.5,4.0], rho in [-0.95,-0.1], H in [0.025,0.5]
    Chosen by hand; the network is only validated within these ranges, and market parameters outside these ranges would require retraining.
  • MC training path count and grid = 60,000 paths; 11 strikes by 8 maturities
    Chosen by hand; affects label accuracy and the network output dimension.
  • Network architecture and optimization hyperparameters = 3 hidden layers of 30 nodes, ELU activation, Adam optimizer
    Chosen by hand without a systematic search; performance may vary with these choices.
assumptions (6)
  • domain assumption Algorithm 3.5 in [34] provides sufficiently accurate approximate prices for the rough Bergomi model.
    Training labels and synthetic validation data are generated by this Monte Carlo scheme; the paper does not quantify its bias or convergence error.
  • domain assumption The implied volatility surface is smooth enough for spline interpolation between grid points.
    The grid-based approach relies on manual interpolation in T and k, justified by smoothness assumptions in Section 3.2.1.
  • domain assumption The chosen uniform parameter ranges cover realistic market scenarios.
    The network is only trained on these ranges; the market calibration in Section 5.3 lies within them.
  • standard math Multilayer neural networks can approximate the pricing map.
    Relies on universal approximation results by Hornik et al. [31, 32]; not proved in this paper.
  • domain assumption The Bayesian regression error is normally distributed.
    Remark 7 acknowledges that this mirrors the sum-of-squares loss; it is not empirically validated.
  • domain assumption The market implied volatility surface is generated by the rough Bergomi model up to liquidity noise.
    This is the standard calibration premise; the paper does not test model adequacy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On deep calibration of (rough) stochastic volatility models." pith.science (2026). https://pith.science/paper/CHON4WMJ

@misc{pith2026190808806,
  author       = {Pith},
  title        = {Pith review of: On deep calibration of (rough) stochastic volatility models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CHON4WMJ}},
  note         = {Machine review of arXiv:1908.08806}
}
read the original abstract

Techniques from deep learning play a more and more important role for the important task of calibration of financial models. The pioneering paper by Hernandez [Risk, 2017] was a catalyst for resurfacing interest in research in this area. In this paper we advocate an alternative (two-step) approach using deep learning techniques solely to learn the pricing map -- from model parameters to prices or implied volatilities -- rather than directly the calibrated model parameters as a function of observed market data. Having a fast and accurate neural-network-based approximating pricing map (first step), we can then (second step) use traditional model calibration algorithms. In this work we showcase a direct comparison of different potential approaches to the learning stage and present algorithms that provide a suffcient accuracy for practical use. We provide a first neural network-based calibration method for rough volatility models for which calibration can be done on the y. We demonstrate the method via a hands-on calibration engine on the rough Bergomi model, for which classical calibration techniques are diffcult to apply due to the high cost of all known numerical pricing methods. Furthermore, we display and compare different types of sampling and training methods and elaborate on their advantages under different objectives. As a further application we use the fast pricing method for a Bayesian analysis of the calibrated model.

Figures

Figures reproduced from arXiv: 1908.08806 by the authors.

Figure 1
Figure 1. SPX Market Implied Volatility surface on 15th February 2018. IVs have been inverted from SPX Weekly European plain vanilla call mid prices and the interpolation is a (non￾arbitrage-free) Delaunay triangulation. Axes denote log-moneyness m = log(K/S0) for strike K and spot S0, time to maturity T in years and market implied volatility σiv(m, T). 1 Introduction Almost half a century after its publication, the option pr… view at source ↗
Figure 2
Figure 2. We compare surface relative errors of the neural network approximator against the [PITH_FULL_IMAGE:figures/full_fig_p019_2.png] view at source ↗
Figure 3
Figure 3. The Figure illustrates the distribution of the approximation error in space after the [PITH_FULL_IMAGE:figures/full_fig_p020_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Calibration relative error per parameter in the test set in the rough Bergomi model [PITH_FULL_IMAGE:figures/full_fig_p021_4.png]
Figure 5
Figure 5. Figure 5: Cumulative Distribution Function (CDF) of Rough Bergomi parameter relative errors [PITH_FULL_IMAGE:figures/full_fig_p022_5.png]
Figure 6
Figure 6. Figure 6: Bayes calibration against synthetic implied volatility surface computed for model param [PITH_FULL_IMAGE:figures/full_fig_p023_6.png]
Figure 7
Figure 7. Figure 7: Liquidity-weighted Bayes calibration against SPX market implied volatility surface from [PITH_FULL_IMAGE:figures/full_fig_p024_7.png]
Figure 8
Figure 8. Figure 8: Out of sample relative errors per parameter calibration [PITH_FULL_IMAGE:figures/full_fig_p026_8.png]
Figure 9
Figure 9. Figure 9: Stars represent the out of sample RMSE via neural network (NN) and brute force Monte [PITH_FULL_IMAGE:figures/full_fig_p027_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Signature-based identification of volatility models from path geometry

    q-fin.CP 2026-07 conditional novelty 5.0 of 10

    Truncated path signatures combined with XGBoost classify stochastic volatility model classes from simulated paths with high accuracy, including under random parameter variation and closely spaced Hurst parameters.

  2. Deep learning interpretability for rough volatility

    q-fin.CP 2024-11 conditional novelty 5.0 of 10

    A neural network trained to invert rough Heston parameters from implied volatility surfaces relies most on short-maturity deep in-the-money prices, a pattern absent in the standard Heston model.

Reference graph

Works this paper leans on

53 extracted references · 47 canonical work pages · cited by 2 Pith papers

  1. [34]

    Functional central limit theorems for rough volatility

    B. Horvath, A. Jacquier and A. Muguruza. Functional central limit theorems for rough volatil- ity. arXiv:1711.03078, 2017

    work page Pith review arXiv 2017

  2. [53]

    H. Stone. Calibrating rough volatility models: a convolutional neural network approach. Preprint, arXiv:1812.05315, 2018. 32

    work page Pith review arXiv 2018

  3. [1]

    Al` os, J

    E. Al` os, J. Le´ on and J. Vives. On the short-time behavior of the implied volatility for jump- diffusion models with stochastic volatility. Finance and Stochastics, 11(4), 571-589, 2007

    work page 2007

  4. [2]

    Antonov, M

    A. Antonov, M. Konikov, M. Spector. Modern SABR Analytics: Formulas and Insights for Quants, Former Physicists and Mathematicians. SpringerBriefs in Quantitative Finance, 2019

    work page 2019

  5. [3]

    Avellaneda, A

    M. Avellaneda, A. Carelli, F. Stella. Following the Bayes path to option pricing. Journal of Computational Intelligence in Finance , 84, 1999

    work page 1999

  6. [4]

    A regularity structure for rough volatility

    C. Bayer, P. Friz, P. Gassiat, J. Martin and B. Stemper. A regularity structure for rough volatility. arXiv:1710.07481, 2017

    work page Pith review arXiv 2017

  7. [5]

    Bayer, P

    C. Bayer, P. Friz and J. Gatheral. Pricing under rough volatility. Quantitative Finance, 16(6): 1-18, 2016

    work page 2016

  8. [6]

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

    C. Bayer, P. Friz, A. Gulisashvili, B. Horvath and B. Stemper. Short-time near the money skew in rough fractional stochastic volatility models. arXiv:1703.05132, 2017

    work page Pith review arXiv 2017

Show all 53 references
  1. [7]

    Bayer and B

    C. Bayer and B. Stemper. Deep calibration of rough stochastic volatility models. Preprint, arXiv:1810.03399

    arXiv

  2. [8]

    Bennedsen, A

    M. Bennedsen, A. Lunde and M.S. Pakkanen. Hybrid scheme for Brownian semistationary processes. Finance and Stochastics, 21(4): 931-965, 2017

    2017

  3. [9]

    L. Bergomi. Stochastic Volatility Modeling. Chapman & Hall/CRC financial mathematical series. Chapman & Hall/CRC , 2015

    2015

  4. [10]

    B. Chen, C. W. Oosterlee and H. Van Der Weide. Efficient unbiased simulation scheme for the SABR stochastic volatility model, 2011

    2011

  5. [11]

    R. Cont. Model Calibration. Encyclopedia of Quantitative Finance , 5, 2010. DOI:10.1002/9780470061602.eqf08002

    2010 doi

  6. [12]

    Culkin and S

    R. Culkin and S. R. Das Machine Learning in Finance: The Case of Deep Learning for Option Pricing. Journal of Investment Management , github:BlackScholesNN 2017. 29

    2017

  7. [13]

    De Spiegeleer, D

    J. De Spiegeleer, D. Madan, S. Reyners and W. Schoutens. Machine learning for quantitative finance: Fast derivative pricing, hedging and fitting. SSRN:3191050, 2018

    2018

  8. [14]

    Dimitroff, D

    G. Dimitroff, D. R¨ oder and C. P. Fries. Volatility model calibration with convolutional neural networks. Preprint, SSRN:3252432, 2018

    2018

  9. [15]

    Eldan and O

    R. Eldan and O. Shamir. The power of depth for feedforward neural neworks.JMLR: Workshop and Conference Proceedings Vol 49:1-34 , 2016

    2016

  10. [16]

    El Euch and M

    O. El Euch and M. Rosenbaum. Perfect hedging in rough Heston models, to appear in The Annals of Applied Probability, 2018

    2018

  11. [17]

    Forde, H

    M. Forde, H. Zhang. Asymptotics for Rough Stochastic Volatility

  12. [18]

    Friedman, R

    J. Friedman, R. Tibshiran and T. Hastie. The Elements of Statistical Learning. Springer New York Inc, 2001

    2001

  13. [19]

    Ferguson and A

    R. Ferguson and A. D. Green. Deeply learning derivatives. Preprint arXiv:1809.02233, 2018

    2018 arXiv

  14. [20]

    Foreman-Mackey, D

    D. Foreman-Mackey, D. W. Hogg, D. Lang, J. Goodman. emcee: the MCMC hammer, Publi- cations of the Astronomical Society of the Pacific , 125(925), 306, 2013

    2013

  15. [21]

    Foreman-Mackey, corner.py: Scatterplot matrices in Python, The Journal of Open Source Software 24, http://dx.doi.org/10.5281/zenodo.45906, 2016

    D. Foreman-Mackey, corner.py: Scatterplot matrices in Python, The Journal of Open Source Software 24, http://dx.doi.org/10.5281/zenodo.45906, 2016

    2016 doi

  16. [22]

    Fukasawa

    M. Fukasawa. Asymptotic analysis for stochastic volatility: martingale expansion. Finance and Stochastics, 15: 635-654, 2011

    2011

  17. [23]

    Gatheral

    J. Gatheral. The volatility surface: a practitioner’s guide, Wiley, 2011

    2011

  18. [24]

    Gatheral, T

    J. Gatheral, T. Jaisson and M. Rosenbaum. Volatility is rough. Quantitative Finance, 18(6): 933-949, 2018

    2018

  19. [25]

    K. Glau, D. Kressner, and F. Statti. Low-rank tensor approximation for Chebyshev interpola- tion in parametric option pricing. arXiv:1902.04367, 2019

    1902 arXiv

  20. [26]

    A. Green. XVA: Credit, Funding and Capital Valuation Adjustments. Wiley, 2015

    2015

  21. [27]

    Hagan, D

    P. Hagan, D. Kumar, A. Lesniewski, and D. Woodward. Managing smile risk. Wilmott Maga- zine, September issue: 84-108 , 2002

    2002

  22. [28]

    J. Han, A. Jentzen, E. Weinan. Overcoming the curse of dimensionality: Solving high- dimensional partial differential equations using deep learning. PNAS.115(34) 8505-8510, Au- gust 2018

    2018

  23. [29]

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

    1993

  24. [30]

    Hernandez

    A. Hernandez. Model calibration with neural networks. Risk, 2017

    2017

  25. [31]

    Hornik, M

    K. Hornik, M. Stinchcombe, and H. White. Multilayer feedforward networks are universal approximators. Neural Networks, 2(5):359-366, 1989

    1989

  26. [32]

    K. Hornik. M. Stinchcombe and H. White. Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networks. Neural Networks. Vol. 3:11, 1990. 30

    1990

  27. [33]

    Henry-Labord` ere

    P. Henry-Labord` ere. Deep primal-dual algorithm for BSDEs: applications of machine learning to CVA and IM. SSRN:3071506

  28. [35]

    Horvath, A

    B. Horvath, A. Muguruza and T. Mehdi. Deep learning volatility. Available at SSRN 3322085, 2019

    2019

  29. [36]

    Horvath, O

    B. Horvath, O. Reichmann. Dirichlet Forms and Finite Element Methods for the SABR Model. SIAM Journal on Financial Mathematics , p. 716-754 (2), May 2018

    2018

  30. [37]

    Hull and A

    J. Hull and A. White. Pricing interest rate derivatives securities. The Review of Financial Studies)(3): 573-592, 1990

    1990

  31. [38]

    J. M. Hutchinson, A. W. Lo and T. Poggio. A Nonparametric Approach to Pricing and Hedging Derivative Securities Via Learning Networks. The Journal of Finance , 49(3) 851-889. Papers and Proceedings Fifty-Fourth Annual Meeting of the American Finance Association, Boston, Massac...

    1994

  32. [39]

    A. Itkin. To sigmoid-based functional description of the volatility smile. Preprint, arXiv:1407.0256, 2014

    2014 arXiv

  33. [40]

    Ioffe and C

    S. Ioffe and C. Szegedy. Batch normalisation: Accelerating deep network training by reducing internal covariate shift. Preprint, arXiv:1502.03167, 2015

    2015 arXiv

  34. [41]

    Kingman and J

    D.P. Kingman and J. Ba, Adam: A Method for Stochastic Optimization. Conference paper, 3rd International Conference for Learning Representations, 2015

    2015

  35. [42]

    Leitao Rodriguez, L.A

    A. Leitao Rodriguez, L.A. Grzelak and C.W. Oosterlee. On an efficient multiple time step Monte Carlo simulation of the SABR model.Quantitative Finance, 17(10), pp.1549-1565, 2017

    2017

  36. [43]

    Levenberg

    K. Levenberg. A Method for the Solution of Certain Non-Linear Problems in Least Squares. Quarterly of Applied Mathematics . 2: pp. 164-168, 1944

    1944

  37. [44]

    Marquardt

    D. Marquardt. An Algorithm for Least-Squares Estimation of Nonlinear Parameters. SIAM Journal on Applied Mathematics . 11 (2): pp. 431-441,1963.‘

    1963

  38. [45]

    S. Liu, A. Borovykh, L. A. Grzelak, C. W. Oosterlee. A neural network-based framework for financial model calibration. Preprint, arXiv:1904.10523, 2019

    1904 arXiv

  39. [46]

    W. A. McGhee. An artificial neural network representation of the SABR stochastic volatility model. Preprint, SSRN:3288882, 2018

    2018

  40. [47]

    McCrickerd, M

    R. McCrickerd, M. Pakkanen, Turbocharging Monte Carlo pricing for the rough Bergomi model, Quantitative Finance 18(11):1877-1886, 2018

    2018

  41. [48]

    Leitao Rodriguez, A

    A. Leitao Rodriguez, A. Grzelak Lech, Cornelis W. Oosterlee. On a one time-step Monte Carlo simulation approach of the SABR model : Application to European options. Applied Mathematics and Computation , 293 p. 461-479, 2017,

    2017

  42. [49]

    Sabate Vidales, D

    M. Sabate Vidales, D. Siska, L. Szpruch Unbiased deep solvers for parametric PDEs arXiv:1810.05094, 2018. 31

    2018 arXiv

  43. [50]

    Sirignano and K

    J. Sirignano and K. Spiliopoulos. DGM: A deep learning algorithm for solving partial differ- ential equations. Journal of Computational Physics 375(15) 1339-1364, December 2018

    2018

  44. [51]

    Setayeshgar, and H

    L. Setayeshgar, and H. Wang. Large deviations for a feed-forward network,Advances in Applied Probability, 43: 2, pp. 545-571, 2011

    2011

  45. [52]

    Shaham, A

    U. Shaham, A. Cloninger, and R. R. Coifman. Provable approximation properties for deep neural networks. Appl. Comput. Harmon. Anal. , 44(3): 537-557, 2018

    2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.