Towards Causal Market Simulators

Dennis Thumm; Luis Ontaneda Mijares

REVIEW 4 major objections 5 minor 36 references

Towards Causal Market Simulators

T0 review · 4 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read A VAE with a DAG-constrained decoder can estimate counterfactual probabilities in synthetic financial time series with L1 error as low as 0.03.

desk verdict A plausible causal-VAE idea that undermines its own empirical claim by omitting the obvious analytical baseline and reporting no error bars. read the letter →

arxiv 2511.04469 v4 pith:V6WMVXSQ submitted 2025-11-06 cs.LG physics.data-anq-fin.CPstat.MEstat.OT

Dennis Thumm , Luis Ontaneda Mijares This is my paper

classification cs.LGphysics.data-anq-fin.CPstat.MEstat.OT

keywords causalinferencecounterfactualreasoningfinancialtimeseriesmarketsimulationstructuralmodelsvariationalautoencodergenerationWassersteindistance

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

This paper tries to establish that a variational autoencoder whose decoder is constrained by a directed acyclic graph—a time-series neural causal model—can answer counterfactual 'what-if' queries about financial time series, not just generate realistic-looking data. In a controlled setting where the data-generating process is a known linear autoregressive system, the model's estimated counterfactual probabilities land within 0.03–0.10 L1 distance of the analytical ground truth. If the approach generalizes beyond these synthetic systems, it would give risk managers a principled way to run stress tests and scenario analyses on market trajectories that respect the causal links between variables.

What carries the argument

The central object is the TNCM-VAE, a variational autoencoder with three components: an encoder that maps an observed time-series window to a latent 'noise' representation; a DAG-structured decoder that generates each variable from its causal parents (including the previous time step) via a normalizing flow; and a training loss combining a causal (adapted) Wasserstein reconstruction term with a KL prior penalty. Counterfactual queries are answered by the standard three-step recipe—abduction (encode the observed sequence into the latent posterior), action (set the intervened variable), prediction (decode forward)—and the DAG ensures interventions propagate only along the specified causal edge

What would settle it

Run the same counterfactual queries on an AR(1) system where the noise is Student-t or where a hidden confounder drives both X and Y, so the encoder's Gaussian latent posterior is provably misspecified; if the L1 error between estimated and analytical counterfactual probabilities jumps well above 0.1, the causal-validity claim collapses.

Watch

Extended reading notes

Core claim

The paper's central claim is that explicit causal structure in a VAE decoder—implemented as a DAG encoding which variables directly influence which—lets the model perform Pearl-style abduction, action, and prediction to produce counterfactual time series whose event probabilities match the true counterfactual distribution. On two AR(1) systems inspired by mean-reverting processes, with the interventions do(X_t = 0) and do(X_t = -2), the model estimates P(Y_{t+1} > threshold | do(...)) with average L1 errors of 0.064 and 0.058 over five time steps. The authors argue this is the first step toward market simulators that can answer causal counterfactual questions rather than only sampling plausi

Load-bearing premise

The learned latent encoding is assumed to be the actual exogenous noise of the data-generating process, but the paper never bounds the difference between the encoder's posterior and the true noise posterior.

Editorial extensions

If this is right

If the model works beyond the synthetic AR(1) cases, it gives a way to generate counterfactual market trajectories for stress testing, where a regulator can ask 'what if this asset had moved differently?' and get an answer consistent with the learned causal graph.
The DAG constraint trades off some reconstruction fidelity for causal correctness, so applications that need causal-valid scenarios rather than minimal reconstruction error are the natural fit.
The low L1 errors across five time steps indicate the model maintains temporal consistency in counterfactual generation, not just a one-step fix.
The method inherits theoretical counterfactual-consistency guarantees only under identifiability conditions on the learned model, so it sets up a concrete checklist for when such a simulator can be trusted.

Reading between the lines

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

The paper does not verify that the encoder's latent posterior equals the true exogenous-noise posterior; a natural next test is to run the same queries on systems with non-Gaussian or heteroskedastic noise and see whether the L1 error grows, since the abduction step is where causal validity is won or lost.
Because the DAG must be supplied, the approach as written presumes the causal graph is known; in real markets the graph is usually unknown, so coupling this decoder with a causal-discovery step is the obvious extension.
The reported errors are averaged over five time steps on one synthetic configuration; a stronger falsifier would be to vary the autoregressive coefficients and noise scales and confirm the error stays below some threshold, rather than only reporting a best-case range.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Quick take: the paper's core idea is reasonable — a VAE with a DAG-structured decoder for counterfactual time series — but the evidence section is not credible as it stands. The synthetic AR(1) setup is well described and the code link helps, and the authors cite the right prior work. But the loss function in Eq. (2) does not match the text's claim of causal Wasserstein training; the DAG enforcement is never specified; and the experiments compare against nothing. In a linear-Gaussian AR(1) with known coefficients, the exact Pearl-style counterfactual is computable by a Kalman smoother, and any sensible linear estimator would give L1 near zero. Reporting absolute L1 of 0.03–0.10 without that baseline tells us nothing. No error bars either, so we can't tell if that's one lucky run.

The Ctf-consistency theorem from Pan & Bareinboim is cited but the conditions in Theorem A.1 are not checked, and the VAE posterior approximation is used as if it were the exact noise posterior. That's a real gap, though it's secondary to the missing baseline.

What's genuinely new is narrow: the specific TNCM-VAE architecture with an explicit DAG in the decoder and its use for counterfactual probability queries on financial AR series. That's a plausible extension of Time-Causal VAE and NCMs, and I'd like to see it developed. But the paper overclaims in the abstract and conclusion ('superior performance compared to existing methods') when no existing methods are tested.

For a workshop this might pass as a position piece, but as a serious research paper the empirical section needs to be redone: add the analytical baseline, a standard VAE without DAG constraints, and error bars over seeds. If the numbers hold up after that, the result would be interesting. For now, I would not rely on the reported L1 distances.

Worth a reading-group discussion about how easy it is to fool yourself with synthetic ground truth, though. Recommendation for peer review: I'd desk-reject as is, but invite a revision if the authors add the missing baselines.

Referee Report

4 major / 5 minor

Summary. The paper proposes TNCM-VAE, a variational-autoencoder-based neural causal model for time series. It claims to enforce causal structure through a DAG-constrained decoder and to train with the causal Wasserstein distance. Counterfactual queries are generated via Pearl's three-step procedure (abduction, action, prediction), using the VAE encoder to infer exogenous latents. The method is evaluated on two synthetic AR(1) processes inspired by the Ornstein–Uhlenbeck model, where the ground-truth counterfactual probabilities are analytically available. The reported L1 distances between model estimates and analytical ground truth are 0.03–0.10. The paper concludes that TNCM-VAE achieves accurate and 'superior' counterfactual generation for financial stress testing and scenario analysis.

Significance. If the central claims were substantiated, the paper would offer a useful step toward causal market simulators: combining VAEs with SCM-based counterfactual reasoning is timely, and the choice of a controlled linear-Gaussian AR setting with analytical ground truth is a sensible way to validate such a model. The availability of a code repository is also a positive. However, the paper, as written, does not establish the claims. The displayed training loss does not contain the advertised causal Wasserstein term, the empirical section reports no baselines or error bars, the invoked counterfactual-consistency theorem is not verified, and the DAG enforcement mechanism is not specified. These are load-bearing omissions rather than presentation issues.

major comments (4)

[Abstract; §2.1, Eq. (2)] The abstract and §2.1 state that the method 'employs the causal Wasserstein distance for training,' but Eq. (2) is a standard ELBO-type loss consisting of an L1 reconstruction term and a KL regularization term. No Wasserstein distance, bicausal coupling, or transport term appears anywhere in the loss or its description. The central training objective is therefore not the one advertised, and the claimed causal regularization is not defined. The authors must either display the actual loss used or remove the causal-Wasserstein claim.
[§3.1, Tables 2–3; Appendix A.3] The empirical evidence consists solely of absolute L1 distances (0.03–0.10) against analytical ground truth. No baselines are reported. In this linear-Gaussian AR(1) setting, an exact counterfactual estimator obtained by applying Pearl's three-step procedure to the known SCM (with abduction performed by a Kalman smoother) is available and would constitute the natural benchmark; without it, the reported L1 values are uninterpretable. For probabilities near 0.5, a constant predictor can achieve L1 of roughly 0.1–0.25, comparable to the reported values. Moreover, Appendix A.3 claims 'superior causal accuracy compared to baseline methods' and 'temporal consistency ... compared to methods that prioritize reconstruction accuracy alone,' yet no such comparisons appear in any table or figure. The absence of error bars or multiple seeds further means the results could reflect a single favorable r
[§2.2, step 1; Theorem A.1] Counterfactual consistency is asserted by citing Theorem A.1, which requires the learned model M to lie in Omega_I(G) and to match the observed distribution P_M*(X,Y). Neither condition is checked. The abduction step uses the VAE approximate posterior Q_phi(U|V) as if it were the exact exogenous-noise posterior of the SCM; no bound on the variational approximation error is provided. Since counterfactual validity in Pearl's framework depends on exact posterior inference, the presented probability estimates cannot be claimed to be causally valid on the basis of the cited theorem. In this synthetic setting, both conditions can in principle be tested; the paper should do so or provide an approximation-error analysis.
[§2.1, Decoder] The decoder is said to 'enforce a DAG structure' that 'encodes the causal relationships,' but the manuscript does not specify how this is achieved in the neural architecture. No adjacency masking, no acyclicity penalty, no parameterization of the conditional distributions in terms of the DAG is given. Since the DAG constraint is the main claimed causal mechanism, this omission prevents reproduction and makes it impossible to assess whether the model is indeed respecting the causal graph during generation.

minor comments (5)

[Table 1] The notation Y^1_{t+1} and X^1_t in Table 1 is confusing; the superscripts are not defined and do not match the text's Y_{t+1}, X_t notation.
[Figures 4–5] The figures are referenced in §3.1 but are not described in the text; no axis labels or legend details are given, so the reader cannot independently evaluate the 'close alignment' claimed.
[Appendix A.1] The Euler–Maruyama derivation is lengthy and not connected to the experiments or counterfactual analysis beyond motivation. It could be condensed substantially.
[§2.2] The phrase 'bounded transport loss' is mentioned as providing theoretical guarantees, but this object is never defined; if this is the causal Wasserstein distance, it should be introduced in §2.1 and used in Eq. (2).
[§3, Eq. (8)] The generative equations are written without explicit time indices on the noise terms; specifying eta_t ~ N(0,1) and epsilon_t ~ N(0,1) immediately after the equations is fine, but the independence assumption between eta and epsilon should be stated explicitly.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the counterfactual accuracy claim is tested against an independent analytical ground truth, and the cited theoretical supports are external to the authors.

full rationale

The paper's derivation chain is not circular. The model is trained on data generated from the known AR equations (Eq. 3) and evaluated against the analytical counterfactual probabilities of those same equations; the target L1 distances are not used as training inputs or fitted parameters. The counterfactual probabilities P(Y_{t+1} > threshold | do(X_t = x)) are computed from the trained VAE via the abduction-action-prediction procedure (§2.2) and then compared with ground truth, so the reported quantities are not equivalent to the loss function by construction. The theoretical justification — the Ctf-consistency condition (Theorem A.1) and the causal Wasserstein framework — is imported from [25] and [1]/[5], none of which shares authors with the present paper; thus no load-bearing self-citation chain exists. The acknowledged limitations in A.3 (higher reconstruction error, scalability, real-world complexity) and the unverified conditions of Theorem A.1 are genuine validity concerns, as is the absence of a baseline or error bars, but these are not circularity: they do not reduce the prediction to an input by definition. No specific circular step can be exhibited from the paper's own equations.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

No new physical or causal entities are introduced; U is the standard VAE latent/exogenous variable. The main burdens are the assumed known DAG, the unverified use of the variational posterior as an exact abduction distribution, and the unverified applicability of the cited consistency theorem.

free parameters (2)

beta (KL regularization weight) = not reported
Weight on the KL term in Eq. (2); chosen by hand; value not given, affects trade-off between reconstruction and posterior fit.
Model hyperparameters = not reported
Latent dimension, GRU hidden size, RealNVP layer count/depth, learning rate, batch size, epochs are unspecified, so the reported L1 numbers cannot be reproduced or judged.

assumptions (5)

domain assumption The causal graph is known a priori: X_{t-1}->X_t, Y_{t-1}->Y_t, X_{t-1}->Y_t
In synthetic experiments this is true by construction; for real financial markets the graph would be unknown, so the method's usefulness depends on this assumption (Section 2, Figure 2).
ad hoc to paper The VAE posterior Q_phi(U|V) is a valid substitute for the exact SCM exogenous posterior in abduction
Section 2.2 step 1 uses the encoder for abduction without a bound on error; without this, Theorem A.1's conditions are not met.
domain assumption Theorem A.1 from [25] applies to the learned model
Appendix A.2 states the theorem without verifying membership in Omega_I(G) or equality of observational distributions; it is invoked as if satisfied.
standard math Euler-Maruyama with Delta t = 1 maps the OU SDE to the AR(1) test equations
Appendix A.1 derivation is correct for the specified discretization; it is background for the synthetic experiments, not for the method itself.
domain assumption Financial market variables satisfy a structural causal model with acyclic temporal dependencies
Conclusion and abstract claim stress testing; not validated. If markets are non-stationary or have hidden confounders/time-varying graphs, counterfactuals are invalid.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Towards Causal Market Simulators." pith.science (2026). https://pith.science/paper/V6WMVXSQ

@misc{pith2026251104469,
  author       = {Pith},
  title        = {Pith review of: Towards Causal Market Simulators},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/V6WMVXSQ}},
  note         = {Machine review of arXiv:2511.04469}
}

read the original abstract

Market generators using deep generative models have shown promise for synthetic financial data generation, but existing approaches lack causal reasoning capabilities essential for counterfactual analysis and risk assessment. We propose a Time-series Neural Causal Model VAE (TNCM-VAE) that combines variational autoencoders with structural causal models to generate counterfactual financial time series while preserving both temporal dependencies and causal relationships. Our approach enforces causal constraints through directed acyclic graphs in the decoder architecture and employs the causal Wasserstein distance for training. We validate our method on synthetic autoregressive models inspired by the Ornstein-Uhlenbeck process, demonstrating superior performance in counterfactual probability estimation with L1 distances as low as 0.03-0.10 compared to ground truth. The model enables financial stress testing, scenario analysis, and enhanced backtesting by generating plausible counterfactual market trajectories that respect underlying causal mechanisms.

Figures

Figures reproduced from arXiv: 2511.04469 by the authors.

**Figure 2.** Directed acyclic graph (DAG) showing relationships [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 4.** Experiment 1: Probability using [PITH_FULL_IMAGE:figures/full_fig_p003_4.png] view at source ↗

Figures from the paper (1 more)

**Figure 5.** Figure 5: Experiment 2: Probability using 𝑑𝑜 intervention counterfactual financial time series. By enforcing causal constraints through DAG-structured decoders and leveraging causal Wasserstein distances for training, our approach achieves superior counterfactual generation qu…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

36 extracted references · 1 canonical work pages

[1]

Beatrice Acciaio, Stephan Eckstein, and Songyan Hou. 2024. Time-Causal VAE: Robust Financial Time Series Generator. arXiv:2411.02947 [cs.LG] https://arxiv. org/abs/2411.02947

arXiv 2024
[2]

Ioana Bica, Ahmed Alaa, and Mihaela Van Der Schaar. 2020. Time series de- confounder: Estimating treatment effects over time in the presence of hidden confounders. InInternational conference on machine learning. PMLR, 884–895

2020
[3]

Ioana Bica, James Jordon, and Mihaela van der Schaar. 2020. Estimating the effects of continuous-valued interventions using generative adversarial networks. Advances in Neural Information Processing Systems33 (2020), 16434–16445

2020
[4]

2009.Arbitrage theory in continuous time

Tomas Björk. 2009.Arbitrage theory in continuous time. Oxford university press

2009
[5]

Patrick Cheridito and Stephan Eckstein. 2025. Optimal transport and Wasserstein distances for causal models.Bernoulli31, 2 (2025), 1351–1376

2025
[6]

Kyunghyun Cho, Bart van Merriënboer, Dzmitry Bahdanau, and Yoshua Bengio
[7]

Rama Cont. 2001. Empirical properties of asset returns: stylized facts and statisti- cal issues.Quantitative finance1, 2 (2001), 223

2001
[8]

Marcos López de Prado. 2025. AI Challenges in Mathematical Investing. Presen- tation at World AI Conference, Shanghai. www.QuantResearch.org Abu Dhabi Investment Authority & Cornell University

2025

Show all 36 references

[9]

Laurent Dinh, Jascha Sohl-Dickstein, and Samy Bengio. 2017. Density estimation using Real NVP. InInternational Conference on Learning Representations. https: //openreview.net/forum?id=HkpbnH9lx

2017
[10]

Joseph L Doob. 1942. The Brownian movement and stochastic equations.Annals of Mathematics43, 2 (1942), 351–369

1942
[11]

Gelin Gao, Bud Mishra, and Daniele Ramazzotti. 2018. Causal data science for financial stress testing.Journal of computational science26 (2018), 294–304

2018
[12]

Mingming Gong, Kun Zhang, Bernhard Schölkopf, Clark Glymour, and Dacheng Tao. 2017. Causal discovery from temporally aggregated time series. InUn- certainty in artificial intelligence: proceedings of the... conference. Conference on Uncertainty in Artificial Intelligence, Vol....

2017
[13]

Campbell R Harvey and Yan Liu. 2015. Backtesting.The Journal of Portfolio Management42, 1 (2015), 13–28

2015
[14]

Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long short-term memory.Neural computation9, 8 (1997), 1735–1780

1997
[15]

Blanka Horvath, Jonathan Plenk, Milena Vuletić, and Raeid Saqur. 2025. Gen- erative Models in Finance: Market Generators, a Paradigm Shift in Financial Modeling. https://papers.ssrn.com/abstract=5284313

2025
[16]

Diederik P Kingma. 2013. Auto-encoding variational bayes.arXiv preprint arXiv:1312.6114(2013)

2013 arXiv
[17]

Samantha Kleinberg and Bud Mishra. 2009. The Temporal Logic of Causal Struc- tures. InProceedings of the 25th Conference on Uncertainty in Artificial Intelligence (UAI). Montreal, Quebec

2009
[18]

Oleksiy Kondratyev and Christian Schwarz. 2019. The market generator.A vailable at SSRN 3384948(2019)

2019
[19]

Luca Lalor and Anatoliy Swishchuk. 2025. Event-Based Limit Order Book Simulation under a Neural Hawkes Process: Application in Market-Making. arXiv:2502.17417 [q-fin.CP] https://arxiv.org/abs/2502.17417

2025
[20]

Junyi Li, Xintong Wang, Yaoyang Lin, Arunesh Sinha, and Michael Wellman
[21]

Ben A. Marconi. 2025. Time Series Foundation Models for Multivariate Financial Time Series Forecasting. arXiv:2507.07296 [q-fin.GN] https://arxiv.org/abs/2507. 07296

2025 arXiv
[22]

Valentyn Melnychuk, Dennis Frauen, and Stefan Feuerriegel. 2022. Causal trans- former for estimating counterfactual outcomes. InInternational conference on machine learning. PMLR, 15293–15329

2022
[23]

Nicola Muca Cirone, Antonio Orvieto, Benjamin Walker, Cristopher Salvi, and Terry Lyons. 2024. Theoretical foundations of deep selective state-space models. Advances in Neural Information Processing Systems37 (2024), 127226–127272

2024
[24]

Daniel Cunha Oliveira, Yutong Lu, Xi Lin, Mihai Cucuringu, and Andre Fu- jita. 2024. Causality-Inspired Models for Financial Time Series Forecasting. arXiv:2408.09960 [q-fin.CP] https://arxiv.org/abs/2408.09960

2024 arXiv
[25]

Yushu Pan and Elias Bareinboim. 2024. Counterfactual Image Editing. InPro- ceedings of the 41st International Conference on Machine Learning (Proceedings of Machine Learning Research, Vol. 235), Ruslan Salakhutdinov, Zico Kolter, Kather- ine Heller, Adrian Weller, Nuria Oliver...

2024
[26]

2009.Causality

Judea Pearl. 2009.Causality. Cambridge University Press

2009
[27]

Audrey Poinsot, Alessandro Leite, Nicolas Chesneau, Michele Sebag, and Marc Schoenauer. 2024. Learning Structural Causal Models through Deep Generative Models: Methods, Guarantees, and Challenges. InProceedings of the Thirty-Third International Joint Conference on Artificial I...

2024 doi
[28]

Jake Robertson, Arik Reuter, Siyuan Guo, Noah Hollmann, Frank Hutter, and Bernhard Schölkopf. 2025. Do-PFN: In-Context Learning for Causal Effect Estimation. In1st ICML Workshop on Foundation Models for Structured Data. https://openreview.net/forum?id=Jb9yNhfsEM

2025
[29]

Bernhard Schölkopf, Francesco Locatello, Stefan Bauer, Nan Rosemary Ke, Nico Kalchbrenner, Anirudh Goyal, and Yoshua Bengio. 2021. Toward causal repre- sentation learning.Proc. IEEE109, 5 (2021), 612–634

2021
[30]

Alik Sokolov, Fabrizzio Sabelli, Behzad Azadie Faraz, Wuding Li, and Luis Seco
[31]

James H Stock and Mark W Watson. 2001. Vector autoregressions.Journal of Economic perspectives15, 4 (2001), 101–115

2001
[32]

Magnus Wiese, Robert Knobloch, Ralf Korn, and Peter Kretschmer. 2020. Quant GANs: deep generation of financial time series.Quantitative Finance20, 9 (2020), 1419–1440

2020
[33]

Kevin Xia, Yushu Pan, and Elias Bareinboim. 2023. Neural Causal Models for Counterfactual Identification and Estimation. InProceedings of the 11th Interna- tional Conference on Learning Representations. https://openreview.net/forum?id= vouQcZS8KfW A Technical Appendices and Su...

2023
[2014]

InProceedings of SSST-8, Eighth Workshop on Syntax, Semantics and Structure in Statistical Translation

On the Properties of Neural Machine Translation: Encoder–Decoder Ap- proaches. InProceedings of SSST-8, Eighth Workshop on Syntax, Semantics and Structure in Statistical Translation. 103–111
[2020]

InProceedings of the AAAI Conference on Artificial Intelligence, Vol

Generating realistic stock market order streams. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 34. 727–734
[2025]

In Transactions of ADIA Lab: Interdisciplinary Advances in Data and Computational Science

Toward Automating Causal Discovery in Financial Markets and Beyond. In Transactions of ADIA Lab: Interdisciplinary Advances in Data and Computational Science. World Scientific, 359–392

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.

[1] [1]

Beatrice Acciaio, Stephan Eckstein, and Songyan Hou. 2024. Time-Causal VAE: Robust Financial Time Series Generator. arXiv:2411.02947 [cs.LG] https://arxiv. org/abs/2411.02947

arXiv 2024

[2] [2]

Ioana Bica, Ahmed Alaa, and Mihaela Van Der Schaar. 2020. Time series de- confounder: Estimating treatment effects over time in the presence of hidden confounders. InInternational conference on machine learning. PMLR, 884–895

2020

[3] [3]

Ioana Bica, James Jordon, and Mihaela van der Schaar. 2020. Estimating the effects of continuous-valued interventions using generative adversarial networks. Advances in Neural Information Processing Systems33 (2020), 16434–16445

2020

[4] [4]

2009.Arbitrage theory in continuous time

Tomas Björk. 2009.Arbitrage theory in continuous time. Oxford university press

2009

[5] [5]

Patrick Cheridito and Stephan Eckstein. 2025. Optimal transport and Wasserstein distances for causal models.Bernoulli31, 2 (2025), 1351–1376

2025

[6] [6]

Kyunghyun Cho, Bart van Merriënboer, Dzmitry Bahdanau, and Yoshua Bengio

[7] [7]

Rama Cont. 2001. Empirical properties of asset returns: stylized facts and statisti- cal issues.Quantitative finance1, 2 (2001), 223

2001

[8] [8]

Marcos López de Prado. 2025. AI Challenges in Mathematical Investing. Presen- tation at World AI Conference, Shanghai. www.QuantResearch.org Abu Dhabi Investment Authority & Cornell University

2025