pith. sign in

arxiv: 2412.00658 · v2 · submitted 2024-12-01 · 💱 q-fin.ST · stat.CO· stat.ME

Probabilistic Predictions of Option Prices with Modular Approximate Bayesian Inference

Worapree Maneesoonthorn , David T. Frazier , Gael M. Martin This is my paper

Pith reviewed 2026-05-23 08:19 UTC · model grok-4.3

classification 💱 q-fin.ST stat.COstat.ME
keywords approximate Bayesian inferenceoption pricingprobabilistic predictionHeston modelmodular inferencehigh-frequency dataspot returnsfinancial forecasting
0
0 comments X

The pith

Modular approximate Bayesian inference produces probabilistic option price predictions by fusing spot returns, high-frequency data and option prices directly from the pricing model without an explicit likelihood for the options.

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

The paper proposes a new approximate Bayesian framework that integrates daily spot returns, high-frequency spot data, and option prices to generate fast probabilistic predictions of future option prices. It works directly from the theoretical pricing model such as Heston and avoids specifying any statistical model or likelihood for the observed option prices themselves. The method yields accurate predictions in realistic settings and remains robust to unmodeled errors in option prices. Empirical results focus on short-maturity options and highlight the speed of real-time predictive updates.

Core claim

A modular approximate Bayesian inferential framework exploits multiple information sources to enable fast calculation of probabilistic predictions of future option prices. The framework operates directly from the theoretical option pricing model and does not require an explicit statistical model or likelihood for the observed option prices. It produces accurate probabilistic option-price predictions and is robust to the presence of such errors.

What carries the argument

Modular approximate Bayesian inference that combines daily spot returns, high-frequency spot data, and option prices to update predictive distributions based on the Heston model.

If this is right

  • Probabilistic predictions of option prices can be computed and updated in real time as new spot and option data arrive.
  • The predictions stay accurate even when option prices contain errors that are never modeled explicitly.
  • Multiple data sources can be fused without constructing a joint likelihood for the option observations.
  • Empirical accuracy holds for short-maturity options under the Heston specification.

Where Pith is reading between the lines

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

  • The same modular structure could be tested on other pricing models such as stochastic volatility with jumps if the data-fusion step generalizes.
  • Real-time predictive distributions might support applications like dynamic hedging where rapid updates matter more than full error modeling.
  • If the robustness to pricing errors persists, the approach could reduce the modeling burden in markets where option data are noisy.

Load-bearing premise

The Heston model is assumed to be close enough to reality that its direct predictions remain useful even when option prices contain unmodeled errors, and the modular procedure fuses the three data sources without systematic bias.

What would settle it

In out-of-sample tests on short-maturity options, check whether the predictive intervals achieve the claimed coverage rates for realized prices and whether adding option data measurably improves accuracy over spot data alone; failure on either count would falsify the claim.

Figures

Figures reproduced from arXiv: 2412.00658 by David T. Frazier, Gael M. Martin, Worapree Maneesoonthorn.

Figure 1
Figure 1. Figure 1: DAG representation of the relationship between the three information sources and the SV [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: An example of soft cutting information modularization in the context of an SV model with [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Proposed structure for the application of modular ABI to the Heston (1993) option [PITH_FULL_IMAGE:figures/full_fig_p019_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Cut posterior for the structural Heston parameters. The approximate posteriors from [PITH_FULL_IMAGE:figures/full_fig_p020_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Number of active call option contracts per trading day that have trade volume larger [PITH_FULL_IMAGE:figures/full_fig_p025_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Posterior median of the Heston model parameters, using the fixed rolling-window esti [PITH_FULL_IMAGE:figures/full_fig_p027_6.png] view at source ↗
read the original abstract

A new approximate Bayesian inferential framework is proposed that exploits multiple information sources -- daily spot returns, high-frequency spot data and option prices -- and enables fast calculation of probabilistic predictions of future option prices. This approach operates directly from the theoretical option pricing model, and does not require an explicit statistical model, or likelihood, for the observed option prices. We demonstrate that our approach produces accurate probabilistic option-price predictions in realistic scenarios and, despite not explicitly modelling option-pricing errors via a statistical model, the method is shown to be robust to the presence of such errors. Predictive accuracy based on the Heston option pricing model is illustrated empirically for short-maturity options, with the rapidity of real-time updates of the predictive distributions highlighted.

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 / 3 minor

Summary. The manuscript proposes a modular approximate Bayesian inference (MABI) framework that fuses daily spot returns, high-frequency intraday spot data, and option prices to generate probabilistic forecasts of future option prices. The method is applied directly to the Heston stochastic volatility model without specifying an explicit likelihood for the option observations. The central claims are that the resulting predictive distributions are accurate in realistic settings and remain robust to unmodeled option-pricing errors, with supporting empirical results shown for short-maturity options and emphasis on computational speed for real-time updating.

Significance. If the empirical and theoretical claims hold, the work would provide a practical route to uncertainty-quantified option-price forecasts that exploits heterogeneous data sources while avoiding the computational cost of a full joint likelihood for noisy option data. The modular construction and direct use of the pricing model are potentially valuable contributions to quantitative finance, particularly if the robustness to pricing errors is convincingly demonstrated beyond the Heston short-maturity case.

major comments (2)
  1. [§4.3, Eq. (15)] §4.3, Eq. (15): the modular posterior combination is presented as bias-free, yet the weighting parameter λ that balances the high-frequency module against the daily-return module is selected by cross-validation on the same option data used for evaluation; this introduces a potential circularity that undermines the claim of operating 'directly from the theoretical model' without statistical modeling of options.
  2. [Table 4] Table 4, rows for 5- and 10-day horizons: the reported 90% predictive interval coverage for out-of-sample options is 0.82–0.87, which is materially below nominal; the paper does not provide a formal test or adjustment for this under-coverage, weakening the robustness conclusion when option errors are present.
minor comments (3)
  1. [§3.2] The definition of the high-frequency module in §3.2 relies on realized volatility but does not cite or compare against standard realized-volatility estimators (e.g., Barndorff-Nielsen & Shephard); adding this reference would clarify the contribution.
  2. [Figure 5] Figure 5 caption states 'posterior predictive densities' but the plotted quantities are actually posterior predictive intervals; correcting the caption would improve clarity.
  3. The abstract claims 'fast calculation' but no wall-clock timings or scaling plots versus MCMC are supplied; a brief computational comparison would support the practicality claim.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback. We address each major comment below, acknowledging valid concerns where they arise and outlining specific revisions to strengthen the manuscript.

read point-by-point responses

  1. Referee: [§4.3, Eq. (15)] §4.3, Eq. (15): the modular posterior combination is presented as bias-free, yet the weighting parameter λ that balances the high-frequency module against the daily-return module is selected by cross-validation on the same option data used for evaluation; this introduces a potential circularity that undermines the claim of operating 'directly from the theoretical model' without statistical modeling of options.

    Authors: We agree that using cross-validation on the option data for selecting λ introduces a form of data-dependent tuning that qualifies as a limited statistical procedure on the option observations. This does create a potential circularity with the evaluation data and tempers the claim of operating entirely without statistical modeling of options. In the revision we will (i) explicitly state that λ is a hyperparameter tuned via CV, (ii) ensure and document that the CV folds are strictly disjoint from the final test sets used for evaluation, and (iii) qualify the “directly from the theoretical model” phrasing to note that only the core posterior combination step avoids an explicit option likelihood, while hyperparameter selection remains data-driven. These clarifications will be added to §4.3 and the abstract. revision: yes

  2. Referee: [Table 4] Table 4, rows for 5- and 10-day horizons: the reported 90% predictive interval coverage for out-of-sample options is 0.82–0.87, which is materially below nominal; the paper does not provide a formal test or adjustment for this under-coverage, weakening the robustness conclusion when option errors are present.

    Authors: The reported coverages of 0.82–0.87 at the 5- and 10-day horizons are indeed below the nominal 90 % level. We will add a formal binomial test (or Wilson-score interval) for the coverage probability in the revised Table 4 and accompanying text. We will also discuss possible sources of under-coverage, including the accumulation of model misspecification over longer horizons and the effect of unmodeled option errors, while noting that the modular approach still yields better-calibrated intervals than the daily-return-only baseline. No post-hoc adjustment to the intervals themselves is planned, as that would alter the predictive procedure; instead we will emphasize the empirical robustness relative to alternatives. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper's central claim is that a modular approximate Bayesian procedure fuses spot returns, high-frequency data and option prices to yield probabilistic option-price predictions directly from the Heston model, without an explicit likelihood for the option observations. No equation or step in the supplied abstract reduces a prediction to a fitted parameter by construction, nor does any load-bearing premise rest on a self-citation whose content is itself unverified. The method is presented as operating from the theoretical pricing model with empirical robustness checks; the derivation therefore remains self-contained against external benchmarks and receives the default non-circularity finding.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no information on free parameters, axioms, or invented entities; all fields left empty.

pith-pipeline@v0.9.0 · 5654 in / 1192 out tokens · 16300 ms · 2026-05-23T08:19:48.299946+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

55 extracted references · 55 canonical work pages

  1. [1]

    A \" t-Sahalia, Y., Li, C., and Li, C. X. (2021a). Closed-form implied volatility surfaces for stochastic volatility models with jumps. Journal of Econometrics , 222(1):364--392

    work page

  2. [2]

    A \" t-Sahalia, Y., Li, C., and Li, C. X. (2021b). Implied stochastic volatility models. The Review of Financial Studies , 34(1):394--450

    work page

  3. [3]

    Almeida, C., Fan, J., Freire, G., and Tang, F. (2023). Can a machine correct option pricing models? Journal of Business & Economic Statistics , 41(3):995--1009

    work page 2023

  4. [4]

    G., Bollerslev, T., Diebold, F

    Andersen, T. G., Bollerslev, T., Diebold, F. X., and Labys, P. (2001). The distribution of realized exchange rate volatility. Journal of the American Statistical Association , 96(453):42--55

    work page 2001

  5. [5]

    Bakshi, G., Cao, C., and Chen, Z. (1997). Empirical performance of alternative option pricing models. The Journal of Finance , 52(5):2003--2049

    work page 1997

  6. [6]

    Barndorff-Nielsen, O. E. and Shephard, N. (2002). Econometric analysis of realized volatility and its use in estimating stochastic volatility models. Journal of the Royal Statistical Society Series B: Statistical Methodology , 64(2):253--280

    work page 2002

  7. [7]

    Barndorff-Nielsen, O. E. and Shephard, N. (2004). Power and bipower variation with stochastic volatility and jumps. Journal of Financial Econometrics , 2(1):1--37

    work page 2004

  8. [8]

    Barndorff-Nielsen, O. E. and Shephard, N. (2006). Econometrics of testing for jumps in financial economics using bipower variation. Journal of Financial Econometrics , 4(1):1--30

    work page 2006

  9. [9]

    Bates, D. S. (1996). Jumps and stochastic volatility: Exchange rate processes implicit in D eutsche mark options. The Review of Financial Studies , 9(1):69--107

    work page 1996

  10. [10]

    and Scholes, M

    Black, F. and Scholes, M. (1973). The pricing of options and corporate liabilities. Journal of Political Economy , 81(3):637--654

    work page 1973

  11. [11]

    G., Nunes, M

    Blum, M. G., Nunes, M. A., Prangle, D., and Sisson, S. A. (2013). A comparative review of dimension reduction methods in approximate B ayesian computation. Statistical Science , 28(2)

    work page 2013

  12. [12]

    Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics , 31(3):307--327

    work page 1986

  13. [13]

    Bollerslev, T., Gibson, M., and Zhou, H. (2011). Dynamic estimation of volatility risk premia and investor risk aversion from option-implied and realized volatilities. Journal of Econometrics , 160(1):235--245

    work page 2011

  14. [14]

    J., and Quaedvlieg, R

    Bollerslev, T., Patton, A. J., and Quaedvlieg, R. (2016). Exploiting the errors: A simple approach for improved volatility forecasting. Journal of Econometrics , 192(1):1--18

    work page 2016

  15. [15]

    A., Keane, M

    Bruins, M., Duffy, J. A., Keane, M. P., and Smith Jr, A. A. (2018). Generalized indirect inference for discrete choice models. Journal of Econometrics , 205(1):177--203

    work page 2018

  16. [16]

    J., Drovandi, C

    Chakraborty, A., Nott, D. J., Drovandi, C. C., Frazier, D. T., and Sisson, S. A. (2023). Modularized B ayesian analyses and cutting feedback in likelihood-free inference. Statistics and Computing , 33(1):33

    work page 2023

  17. [17]

    T., and Renault, E

    Chaudhuri, S., Frazier, D. T., and Renault, E. (2018). Indirect inference with endogenously missing exogenous variables. Journal of Econometrics , 205(1):55--75

    work page 2018

  18. [18]

    Christoffersen, P., Jacobs, K., Ornthanalai, C., and Wang, Y. (2008). Option valuation with long-run and short-run volatility components. Journal of Financial Economics , 90(3):272--297

    work page 2008

  19. [19]

    Corsi, F. (2009). A simple approximate long-memory model of realized volatility. Journal of Financial Econometrics , 7(2):174--196

    work page 2009

  20. [20]

    Corsi, F., Mittnik, S., Pigorsch, C., and Pigorsch, U. (2008). The volatility of realized volatility. Econometric Reviews , 27(1-3):46--78

    work page 2008

  21. [21]

    Das, S. R. and Sundaram, R. K. (1999). Of smiles and smirks: A term structure perspective. Journal of Financial and Quantitative Analysis , 34(2):211--239

    work page 1999

  22. [22]

    Engle, R. F. (1982). Autoregressive conditional heteroscedasticity with estimates of the variance of U nited K ingdom inflation. Econometrica: Journal of the Econometric Society , pages 987--1007

    work page 1982

  23. [23]

    Eraker, B. (2004). Do stock prices and volatility jump? R econciling evidence from spot and option prices. The Journal of Finance , 59(3):1367--1403

    work page 2004

  24. [24]

    and Okou, C

    Feunou, B. and Okou, C. (2019). Good volatility, bad volatility, and option pricing. Journal of Financial and Quantitative Analysis , 54(2):695--727

    work page 2019

  25. [25]

    S., Martin, G

    Forbes, C. S., Martin, G. M., and Wright, J. (2007). Inference for a class of stochastic volatility models using option and spot prices: Application of a bivariate K alman filter. Econometric Reviews , 26(2-4):387--418

    work page 2007

  26. [26]

    Frazier, D. T. and Koo, B. (2021). Indirect inference for locally stationary models. Journal of Econometrics , 223(1):1--27

    work page 2021

  27. [27]

    T., Maneesoonthorn, W., Martin, G

    Frazier, D. T., Maneesoonthorn, W., Martin, G. M., and McCabe, B. P. (2019a). Approximate B ayesian forecasting. International Journal of Forecasting , 35(2):521--539

    work page

  28. [28]

    T., Martin, G

    Frazier, D. T., Martin, G. M., Robert, C. P., and Rousseau, J. (2018). Asymptotic properties of approximate B ayesian computation. Biometrika , 105(3):593--607

    work page 2018

  29. [29]

    T., Oka, T., and Zhu, D

    Frazier, D. T., Oka, T., and Zhu, D. (2019b). Indirect inference with a non-smooth criterion function. Journal of Econometrics , 212(2):623--645

    work page

  30. [30]

    Fulop, A., Li, J., and Yu, J. (2015). Self-exciting jumps, learning, and asset pricing implications. The Review of Financial Studies , 28(3):876--912

    work page 2015

  31. [31]

    Gospodinov, N., Komunjer, I., and Ng, S. (2017). Simulated minimum distance estimation of dynamic models with errors-in-variables. Journal of Econometrics , 200(2):181--193

    work page 2017

  32. [32]

    Gourieroux, C., Monfort, A., and Renault, E. (1993). Indirect inference. Journal of Applied Econometrics , 8(S1):S85--S118

    work page 1993

  33. [33]

    Heston, S. L. (1993). 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

    work page 1993

  34. [34]

    and Neftci, S

    Hirsa, A. and Neftci, S. N. (2013). An I ntroduction to the M athematics of F inancial D erivatives . Academic press

    work page 2013

  35. [35]

    and Tauchen, G

    Huang, X. and Tauchen, G. (2005). The relative contribution of jumps to total price variance. Journal of Financial Econometrics , 3(4):456--499

    work page 2005

  36. [36]

    Jiang, G. J. (2002). Testing option pricing models with stochastic volatility, random jumps and stochastic interest rates. International Review of Finance , 3(3-4):233--272

    work page 2002

  37. [37]

    Jiang, G. J. and Tian, Y. S. (2005). The model-free implied volatility and its information content. The Review of Financial Studies , 18(4):1305--1342

    work page 2005

  38. [38]

    Jones, C. S. (2003). The dynamics of stochastic volatility: evidence from underlying and options markets. Journal of Econometrics , 116(1-2):181--224

    work page 2003

  39. [39]

    Kim, S., Shephard, N., and Chib, S. (1998). Stochastic volatility: likelihood inference and comparison with ARCH models. The Review of Economic Studies , 65(3):361--393

    work page 1998

  40. [40]

    Koopman, S. J. and Scharth, M. (2012). The analysis of stochastic volatility in the presence of daily realized measures. Journal of Financial Econometrics , 11(1):76--115

    work page 2012

  41. [41]

    M., and Martin, V

    Lim, G.-C., Martin, G. M., and Martin, V. L. (2005). Parametric pricing of higher order moments in S&P500 options. Journal of Applied Econometrics , 20(3):377--404

    work page 2005

  42. [42]

    S., and Martin, G

    Maneesoonthorn, W., Forbes, C. S., and Martin, G. M. (2017). Inference on self-exciting jumps in prices and volatility using high-frequency measures. Journal of Applied Econometrics , 32(3):504--532

    work page 2017

  43. [43]

    M., and Forbes, C

    Maneesoonthorn, W., Martin, G. M., and Forbes, C. S. (2020). High-frequency jump tests: Which test should we use? Journal of Econometrics , 219(2):478--487

    work page 2020

  44. [44]

    M., Forbes, C

    Maneesoonthorn, W., Martin, G. M., Forbes, C. S., and Grose, S. D. (2012). Probabilistic forecasts of volatility and its risk premia. Journal of Econometrics , 171(2):217--236

    work page 2012

  45. [45]

    M., Forbes, C

    Martin, G. M., Forbes, C. S., and Martin, V. L. (2005). Implicit B ayesian inference using option prices. Journal of Time Series Analysis , 26(3):437--462

    work page 2005

  46. [46]

    M., Frazier, D

    Martin, G. M., Frazier, D. T., and Robert, C. P. (2024). Approximating B ayes in the 21st century. Statistical Science , 39(1):20--45

    work page 2024

  47. [47]

    M., McCabe, B

    Martin, G. M., McCabe, B. P., Frazier, D. T., Maneesoonthorn, W., and Robert, C. P. (2019). Auxiliary likelihood-based approximate B ayesian computation in state space models. Journal of Computational and Graphical Statistics , 28(3):508--522

    work page 2019

  48. [48]

    and Medeiros, M

    McAleer, M. and Medeiros, M. C. (2008). Realized volatility: A review. Econometric Reviews , 27(1-3):10--45

    work page 2008

  49. [49]

    J., Drovandi, C., and Frazier, D

    Nott, D. J., Drovandi, C., and Frazier, D. T. (2023). Bayesian inference for misspecified generative models. Annual Review of Statistics and Its Application , 11

    work page 2023

  50. [50]

    Pan, J. (2002). The jump-risk premia implicit in options: Evidence from an integrated time-series study. Journal of Financial Economics , 63(1):3--50

    work page 2002

  51. [51]

    Ruiz, E. (1994). Quasi-maximum likelihood estimation of stochastic volatility models. Journal of Econometrics , 63(1):289--306

    work page 1994

  52. [52]

    A., Fan, Y., and Beaumont, M

    Sisson, S. A., Fan, Y., and Beaumont, M. (2018). Handbook of Approximate B ayesian Computation . CRC Press

    work page 2018

  53. [53]

    S., Yu, W., Nott, D

    Smith, M. S., Yu, W., Nott, D. J., and Frazier, D. (2023). Cutting feedback in misspecified copula models. arXiv preprint arXiv:2310.03521

    work page arXiv 2023

  54. [54]

    and Swamy, M

    Thulasiraman, K. and Swamy, M. N. (2011). Graphs: T heory and A lgorithms . John Wiley & Sons

    work page 2011

  55. [55]

    West, M. (1993). Approximating posterior distributions by mixtures. Journal of the Royal Statistical Society: Series B (Methodological) , 55(2):409--422

    work page 1993