Pith. sign in

REVIEW 5 major objections 5 minor 34 references

Deep Learning Meets Queue-Reactive: A Framework for Realistic Limit Order Book Simulation

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read MDQR, a neural point-process model acting on the whole limit order book, generates Bund-futures dynamics that reproduce the square-root impact law, cross-queue correlations, and order-size distributions.

desk verdict A useful neural QR extension for LOB simulation that reproduces several emergent stylized facts on Bund futures, but the validation needs a real holdout and some cleanup before the stronger claims hold. read the letter →

arxiv 2501.08822 v1 pith:U2T63HKC submitted 2025-01-15 q-fin.TR cs.LG

Hamza Bodor , Laurent Carlier This is my paper · ORCID
classification q-fin.TRcs.LG
keywords limitorderbooksimulationQueue-Reactivemodelmarketimpactsquare-rootlawsizedistributionmicrostructureneuralpointprocessBundfutures
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 claims that the Multidimensional Deep Queue-Reactive (MDQR) model, a neural-network generalization of the Queue-Reactive point-process framework, can generate realistic limit order book dynamics for the Bund futures market. MDQR relaxes the earlier assumption that queues at different price levels behave independently, conditions order-size probabilities on event type, price level, and market state, and enriches the input state with spread, trade imbalance, and past-event information. Against historical Bund data, MDQR is shown to reproduce the square-root law of market impact (fitted exponent 0.55), the negative correlation between best bid and ask queue sizes, the stationary and conditional distributions of order sizes, and heavy-tailed one-minute returns. If correct, the model gives practitioners a fast event-by-event simulator, about 0.037 ms per event, suitable for reinforcement learning and realistic backtesting.

What carries the argument

The central object is the multidimensional marked point-process intensity: for each event, an intensity lambda_{eta,ell}(x;theta) for event type eta and price level ell, together with a conditional order-size distribution p(s | eta, ell, x;theta). The neural network is the mechanism that lets these quantities depend on the joint state of the whole book rather than on one queue's size, which is what carries the cross-level dependency. The factorization of the joint likelihood into an arrival-intensity term and a size term keeps the interpretable Poisson structure for event timing and type while letting a separate classifier handle order sizes, and the network architecture is a two-hidden-layer MLP, dimension 256 by 64, with output dimensions 30 for intensities and 200 for the softmax size distribution.

What would settle it

Take the first two months of the Bund dataset, calibrate MDQR with the same 80/20 early-stopping split, then generate simulated books for a later month that was not used at all; if the fitted impact exponent, the -0.54 best bid-ask correlation, and the order-size quantiles degrade toward QR or SAQR values in that out-of-sample period, the claim that MDQR captures real market dynamics is falsified.

Watch

Extended reading notes

Core claim

MDQR factorizes the likelihood of an order book event sequence into an arrival-intensity term and an order-size term. The arrival term is a marked conditional Poisson process in which a neural network maps a 25-dimensional state (queue sizes across five bid and five ask levels, spread, trade imbalance over four horizons, and embeddings of last-event types and hour) to intensities for 30 possible event-level combinations; the size term is a categorical distribution over 200 order-size classes output by a separate network. The central discovery is that this multidimensional extension, unlike QR and SAQR, reproduces the empirical cross-queue correlation structure, including the -0.54 best bid-ask correlation, and a concave-then-relaxing market impact profile whose maximum impact scales as quantity to the 0.55 power with R-squared 0.89. The same model also outperforms the QR, SAQR, and DeepLOB benchmarks in classifying short-horizon mid-price moves, reaching balanced accuracy 0.63 versus 0.54 for DeepLOB.

Load-bearing premise

The load-bearing premise is that the matches to historical stylized facts demonstrate genuine market-capturing behavior, but the paper never states that the comparison period was held out from calibration, so the reported agreement could be an in-sample fit.

Editorial extensions

If this is right

  • Event-by-event generation at 0.037 ms per event (about 0.92 minutes per simulated Bund trading day) makes MDQR a practical environment for reinforcement-learning training and realistic backtesting.
  • Reproducing the square-root market impact law (fitted exponent 0.55, R-squared 0.89) means execution simulations should show concave temporary impact and relaxation after a metaorder ends rather than a naive linear impact.
  • Capturing the -0.54 best bid-ask queue correlation and the positive same-side correlations preserves the queue-imbalance signals that many trading strategies depend on.
  • Conditional and stationary order-size distributions that track historical data make the simulator suitable for studying execution-quality questions, as in the paper's fill-ratio analysis where price level drives fill rate with correlation -0.79.

Reading between the lines

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

  • The paper does not isolate the contribution of the size model from the multidimensional intensities; plugging MDQR's conditional size sampler into plain QR or SAQR would measure how much of the distributional improvement comes from size conditioning alone.
  • The central role of trade-imbalance features suggests that realism is partly carried by hand-crafted state rather than learned representations, so an end-to-end variant that learns state from raw order book snapshots is a natural next comparison.
  • The fill-ratio analysis could become an external validation tool: if simulated fill probabilities by price level, lifetime, and quantity can be compared with observed execution outcomes from a broker, that would test the simulator against a quantity the paper only generates internally.
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

5 major / 5 minor

Summary. The paper proposes the Multidimensional Deep Queue-Reactive (MDQR) model, an extension of the queue-reactive framework of Huang et al. (2015). MDQR replaces hand-specified intensity functions with an MLP that takes as input a multi-level order book state, thereby coupling dynamics across ten price levels, and adds a separate neural-network categorical model for order sizes over 200 classes. The parameters are estimated by maximizing a factored point-process likelihood on three months of Euro-Bund Futures data. The authors demonstrate the model through a series of experiments: market impact profiles and a fitted square-root impact law, event-transition matrices, mid-price movement prediction, queue-size distributions, cross-queue correlations, return distributions, event-count and volume distributions, order-size distributions, fill-ratio analysis, and runtime comparisons. They conclude that MDQR captures key stylized facts while retaining computational efficiency suitable for reinforcement-learning and backtesting applications.

Significance. If the empirical claims hold, the paper offers a practically useful simulator that combines the interpretable point-process foundation of queue-reactive models with the flexibility of deep learning. The strongest positive evidence is that some reproduced properties, such as the negative best-bid/best-ask queue correlation and the concave market-impact profile, are not directly imposed by the training objective and thus appear to be emergent from the fitted state-dependent intensities. The explicit likelihood formulation also gives the model a well-posed calibration target. However, the validation as presented is substantially in-sample, and several load-bearing comparisons lack uncertainty quantification. These issues are fixable and do not negate the value of the framework, but they currently prevent the paper from fully supporting the abstract's causal-sounding claim that the model captures real market dynamics.

major comments (5)
  1. [§3.3, §4.2, §4.5–§4.7] The validation of the central claim is entirely in-sample. Section 3.3 reports an 80/20 data split only for early stopping, and Section 4.2 states that the MDQR data 'follows the framework established in Section 3.2' with a change only in the inter-event time definition. None of the stylized-fact comparisons in Sections 4.5–4.7 specifies a held-out test period; the simulated statistics are compared with historical data from the same calibration window. Because the intensity and size networks are trained by maximum likelihood on these same days, the reported matches for queue-size distributions, return quantiles, and order-size distributions could be in-sample fits rather than evidence of generalization. A temporal holdout (e.g., train on one sub-period and validate on another) and a clear statement of which historical period is used for each comparison are needed before the abstract's claim of capturing key market properties is supported.
  2. [§4.5.1, Fig. 8] The square-root law claim rests on a power-law fit (exponent 0.55, R²=0.89) to maximum impact values simulated by the same calibrated model. No confidence interval is reported for the exponent, and there is no comparison with empirical metaorder impact curves for the Bund contract. The result is therefore an internal consistency check that the simulated impact is concave in inventory, not a demonstration that the model matches the market's actual impact scaling. Report parameter uncertainty, ideally over multiple simulation seeds, or compare with measured impact data if available.
  3. [§4.6, Table 6] The mid-price prediction comparison is not controlled across models. DeepLOB is calibrated and evaluated on an explicit train/validation/test split, while MDQR's predictive accuracy is obtained by simulating forward events from observed order book states; the text does not state whether those states lie outside the MDQR training period. As a result, the superior balanced accuracy of MDQR (0.63 versus 0.54) may reflect in-sample conditioning rather than genuine predictive skill. The same temporal holdout should be used for all models in this comparison.
  4. [§4.7.2, Fig. 12] The cross-queue correlation matrices are shown without confidence intervals or repeated-seed variability, although many entries of the MDQR matrix are close to the empirical values (e.g., 0.28 versus 0.27 at (bid0, bid1); 0.23 versus 0.17 at several off-diagonal positions). A bootstrap over historical days and over simulation seeds would establish whether the reported agreement is statistically meaningful. Similarly, the QQ plots in Figures 10 and 13 lack uncertainty bands, so the visual 'close alignment' claims are not quantified.
  5. [§4.7.5, Table 8] The runtime comparison is not apples-to-apples. MDQR timings are measured on the authors' AMD EPYC server, while the LOBGAN, RNN, and WGAN figures appear to be taken from their original papers, where hardware and implementation details differ. Before claiming superior computational efficiency, the baselines should be run in the same environment, or the comparison should be clearly labeled as indicative rather than benchmark-grade.
minor comments (5)
  1. [§3, Fig. 6] There are typos: 'occurung' should be 'occurring' in Section 3, and the Figure 6 caption contains 'permanant merket impact'.
  2. [§3.4.3, Fig. 3] The left panel is labeled 'Log-likelihood' but the values are positive and the text says higher is better; if the plotted quantity is the negative log-likelihood, the axis should be relabeled.
  3. [§4.4, Table 4] The stated input dimensions (25 for the intensity model, 27 for the size model) do not obviously match the feature list in Table 3, which includes 10 queue sizes, spread, 4 trade imbalances, 10 categorical last-event-type embeddings, and a 9-category hour embedding. Please clarify how the input vector is assembled and how the embeddings are counted.
  4. [§4.5.1, Fig. 8 caption] The caption contains 'confidence itervals', which should be 'confidence intervals'.
  5. [§4.7.6, Fig. 18] The QQ plot for the stationary order-size distribution covers only sizes up to roughly 35 lots, while the density plot extends to 175 lots; please state whether the tail above the displayed QQ range is included in the assessment and how the 200-class discretization treats sizes above 200.

Circularity Check

1 steps flagged · score 6.0 of 10

The order-size distribution 'validation' reduces to the training objective; other headline results are emergent.

  1. fitted input called prediction [Section 4.7.6 'Order Sizes Distribution', Eq. (5), Figures 16-18]
    "The MDQR sizing model appears to capture these distinct conditional distributions, as evidenced by the close alignment between empirical and simulated patterns in both cases."

    The order-size component is trained by minimizing Eq. (5), a cross-entropy loss between the model's categorical size probabilities and the observed one-hot sizes in the same Bund dataset later used for validation. The network's output is therefore a direct estimate of the empirical conditional size distribution p(s_k | eta_k, l_k, x_k). Section 4.7.6 then validates the model by comparing simulated conditional distributions at fixed queue sizes (Figures 16-17) and the stationary order-size distribution (Figure 18) against that same empirical data. The figure caption even labels the MDQR curve 'Fitted distribution'.

full rationale

The paper's other headline validations are not circular by construction. The square-root impact law (Section 4.5.1, exponent 0.55) and the cross-queue correlation matrices (Section 4.7.2) are emergent properties of the fitted simulator: no term in Eq. (4) or Eq. (5) directly targets those statistics, so their reproduction is genuine evidence of model behavior, even though the comparisons are in-sample. The mid-price prediction task (Section 4.6) is also not forced by the loss. Self-citations to Bodor and Carlier (2024) are present for preprocessing and stylized facts, but the load-bearing derivation of MDQR does not reduce to those citations; the intensity and size networks are estimated from data in this paper. The main concern beyond the order-size circularity is that Section 3.3's 80/20 split is described only for early stopping, and Section 4.5-4.7 comparisons are not tied to a held-out period; that is a generalization-validity weakness rather than a construction-level circularity. Because the paper's abstract emphasizes 'particular strength in reproducing both conditional and stationary distributions of order sizes', and that strength reduces to the training objective, the score is 6 rather than lower.

Assumptions & free parameters 7 free parameters · 6 assumptions · 0 invented entities

The central claim rests on the marked point process likelihood, the hand-crafted state vector, the categorical size model, and the empirical preprocessing choices. The model introduces no new physical or market entities. The main free parameters are neural weights and hand-chosen feature/class settings; the main axioms are the Poisson assumption, the likelihood factorization, and the use of empirical queue initialization in simulation.

free parameters (7)
  • Intensity network weights (theta_lambda) = learned (MLP 256-64, output 30)
    Defines all event arrival intensities in Eq. (4); the core of the generative model.
  • Order-size network weights (theta_s) = learned (MLP 256-64, output 200)
    Defines categorical size distribution in Eq. (5); drives the order-size stylized facts.
  • Order-size class count C = 200
    Chosen because over 99.9% of observed sizes are below 200; fixes the resolution and truncates the tail.
  • Trade imbalance horizons H = 20s, 1min, 5min, 15min
    Hand-selected feature windows (Section 4.3); the market impact profile depends on including these features.
  • Queue normalization factors (AES per level) = from Table 1 (e.g., 6.25 lots at level 1)
    Queues are divided by average event size before training (Section 3.2), changing the state representation.
  • Market impact power-law exponent = 0.55 (R^2=0.89)
    Fitted to simulated maximum impact vs inventory in Fig 8; used to support the square-root law claim but is not a model parameter.
  • Architecture hyperparameters = hidden [256,64], embedding dim 2, LR 1e-5 to 1e-3, early stopping 10 epochs
    Chosen by empirical testing (Sections 3.3, 4.4); affect fit but not the conceptual claim.
assumptions (6)
  • domain assumption Conditional Poisson arrival process per event category and price level
    Assumed in Eq. (3) and used to write the likelihood; excludes self-exciting behavior beyond the features in x_k.
  • domain assumption Likelihood factorization into arrival process and size distribution
    Eq. (3) separates timing/category from size; the model trains two networks independently, assuming no dependence between size and inter-arrival time beyond the state.
  • domain assumption Markovian sufficiency of the hand-crafted state x_k
    Section 4.3 defines features (current queues, spread, trade imbalance, last event types, hour); any dependence outside these is unmodeled.
  • domain assumption Empirical queue initialization after reference price changes
    Section 3.2 initializes new price-level queues from empirical distributions, injecting historical information into simulated states.
  • domain assumption Order flow classification methodology from Bodor and Carlier (2024)
    Updates are classified as limit, cancel, or market using the authors' prior work; accepted without independent verification in this paper.
  • standard math Consistency of maximum likelihood for marked point processes
    The negative log-likelihood objective Eq. (4) is assumed to yield consistent intensity estimates, a standard result for Poisson processes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Learning Meets Queue-Reactive: A Framework for Realistic Limit Order Book Simulation." pith.science (2026). https://pith.science/paper/U2T63HKC

@misc{pith2026250108822,
  author       = {Pith},
  title        = {Pith review of: Deep Learning Meets Queue-Reactive: A Framework for Realistic Limit Order Book Simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/U2T63HKC}},
  note         = {Machine review of arXiv:2501.08822}
}
read the original abstract

The Queue-Reactive model introduced by Huang et al. (2015) has become a standard tool for limit order book modeling, widely adopted by both researchers and practitioners for its simplicity and effectiveness. We present the Multidimensional Deep Queue-Reactive (MDQR) model, which extends this framework in three ways: it relaxes the assumption of queue independence, enriches the state space with market features, and models the distribution of order sizes. Through a neural network architecture, the model learns complex dependencies between different price levels and adapts to varying market conditions, while preserving the interpretable point-process foundation of the original framework. Using data from the Bund futures market, we show that MDQR captures key market properties including the square-root law of market impact, cross-queue correlations, and realistic order size patterns. The model demonstrates particular strength in reproducing both conditional and stationary distributions of order sizes, as well as various stylized facts of market microstructure. The model achieves this while maintaining the computational efficiency needed for practical applications such as strategy development through reinforcement learning or realistic backtesting.

Figures

Figures reproduced from arXiv: 2501.08822 by the authors.

Figure 1
Figure 1. Transition matrix of events on simulated markets vs historical data. The rows of the matrix represent the [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Comparison of DQR (with xk = [qk, hk]) and QR trade order arrival intensities across trading hours, averaged over queue sizes. 3.4.3. Impact of Feature Enrichment on Model Performance To further validate the impact of enriching the state space xk, we evaluated the performance of the Deep Queue-Reactive (DQR) model across three key metrics: • Log-likelihood of the model: A higher log-likelihood indicates better align… view at source ↗
Figure 3
Figure 3. Model performance comparison across different feature sets: (1) Vanilla QR model ( [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (18 more)
Figure 4
Figure 4. Figure 4: A conceptual illustration of the transition from a single-queue model (left) to the multidimensional MDQR [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: Training and validation negative log-likelihood loss evolution for the order intensity model. The main plot [PITH_FULL_IMAGE:figures/full_fig_p016_5.png]
Figure 6
Figure 6. Figure 6: Average shape of price impact. During its execution, a buy metaorder drives the price upward, reaching [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]
Figure 7
Figure 7. Figure 7: Market impact profiles for two models: one using only queue sizes and the other incorporating trade [PITH_FULL_IMAGE:figures/full_fig_p017_7.png]
Figure 8
Figure 8. Figure 8: Average price change for different inventories (left) and maximum market impact ( [PITH_FULL_IMAGE:figures/full_fig_p018_8.png]
Figure 9
Figure 9. Figure 9: Transition matrix of events on simulated markets vs historical for events on the best prices. [PITH_FULL_IMAGE:figures/full_fig_p019_9.png]
Figure 10
Figure 10. Figure 10: Comparison of real and simulated order book queue sizes. [PITH_FULL_IMAGE:figures/full_fig_p021_10.png]
Figure 11
Figure 11. Figure 11: Average LOB volumes for the first five levels at the ask side. [PITH_FULL_IMAGE:figures/full_fig_p022_11.png]
Figure 12
Figure 12. Figure 12: Correlation matrices of queue volumes across different price levels (5 bid and 5 ask levels) for historical [PITH_FULL_IMAGE:figures/full_fig_p023_12.png]
Figure 13
Figure 13. Figure 13: Left: Distribution of one-minute returns for real data and various model variants, highlighting differences in return profiles. Right: Q-Q plot comparing model-generated returns with real returns, showing the degree of alignment between simulated and actual data. MDQR…
Figure 14
Figure 14. Figure 14: Distribution of event counts per 5-minute window across different model specifications. Box plots show [PITH_FULL_IMAGE:figures/full_fig_p024_14.png]
Figure 15
Figure 15. Figure 15: Distribution of aggregated volumes per 5-minute window across different model specifications. Box plots [PITH_FULL_IMAGE:figures/full_fig_p025_15.png]
Figure 16
Figure 16. Figure 16: Fitted distribution (left) and the corresponding QQ-plot (right) for trade orders arriving when the size [PITH_FULL_IMAGE:figures/full_fig_p027_16.png]
Figure 17
Figure 17. Figure 17: Fitted distribution (left) and the corresponding QQ-plot (right) for limit orders arriving when the size of [PITH_FULL_IMAGE:figures/full_fig_p027_17.png]
Figure 18
Figure 18. Figure 18: Left: Stationary distribution of order sizes for real data and the MDQR model, illustrating the similarity between the two distributions. Right: Q-Q plot comparing the stationary distribution of order sizes generated by the MDQR model with real data, demonstrating the…
Figure 19
Figure 19. Figure 19: Distribution of fill ratios across different parameter values. Box plots show the quartiles, whiskers extend to [PITH_FULL_IMAGE:figures/full_fig_p029_19.png]
Figure 20
Figure 20. Figure 20: Heatmaps showing the mean fill ratio for different parameter combinations. Left: Quantity vs Period; [PITH_FULL_IMAGE:figures/full_fig_p030_20.png]
Figure 21
Figure 21. Figure 21: Feature importance analysis using two different methods. Left: Random Forest feature importance, [PITH_FULL_IMAGE:figures/full_fig_p030_21.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 28 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  3. [3]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  4. [4]

    Huang , author C.-A

    author W. Huang , author C.-A. Lehalle , author M. Rosenbaum , title Simulating and analyzing order book data: The queue-reactive model , journal Journal of the American Statistical Association volume 110 ( year 2015 ) pages 107--122

    work page 2015

  5. [5]

    author M. D. Gould , author M. A. Porter , author S. Williams , author M. McDonald , author D. J. Fenn , author S. D. Howison , title Limit order books , journal Quantitative Finance volume 13 ( year 2013 ) pages 1709--1742

    work page 2013

  6. [6]

    Cont , title Statistical modeling of high-frequency financial data , journal IEEE Signal Processing Magazine volume 28 ( year 2011 ) pages 16--25

    author R. Cont , title Statistical modeling of high-frequency financial data , journal IEEE Signal Processing Magazine volume 28 ( year 2011 ) pages 16--25

    work page 2011

  7. [7]

    Cartea , author S

    author \'A . Cartea , author S. Jaimungal , author J. Penalva , title Algorithmic and high-frequency trading , publisher Cambridge University Press , year 2015

    work page 2015

  8. [8]

    Kirilenko , author A

    author A. Kirilenko , author A. S. Kyle , author M. Samadi , author T. Tuzun , title The flash crash: High-frequency trading in an electronic market , journal The Journal of Finance volume 72 ( year 2017 ) pages 967--998

    work page 2017

Show all 34 references
  1. [9]

    Gatheral , title No-dynamic-arbitrage and market impact , journal Quantitative Finance volume 10 ( year 2010 ) pages 749--759

    author J. Gatheral , title No-dynamic-arbitrage and market impact , journal Quantitative Finance volume 10 ( year 2010 ) pages 749--759

    2010

  2. [10]

    Cont , author A

    author R. Cont , author A. De Larrard , title Price dynamics in a markovian limit order market , journal SIAM Journal on Financial Mathematics volume 4 ( year 2014 ) pages 1--25

    2014

  3. [11]

    Bodor , author L

    author H. Bodor , author L. Carlier , title A novel approach to queue-reactive models: The importance of order sizes , journal arXiv preprint arXiv:2405.18594 ( year 2024 )

    2024 arXiv

  4. [12]

    Sirignano , author R

    author J. Sirignano , author R. Cont , title Universal features of price formation in financial markets: perspectives from deep learning , journal Quantitative Finance volume 19 ( year 2019 ) pages 1449--1459

    2019

  5. [13]

    Bouchaud , author M

    author J.-P. Bouchaud , author M. M \'e zard , author M. Potters , et al., title Statistical properties of stock order books: empirical results and models , journal Quantitative Finance volume 2 ( year 2002 ) pages 251--256

    2002

  6. [14]

    Potters , author J.-P

    author M. Potters , author J.-P. Bouchaud , title More statistical properties of order books and price impact , journal Physica A: Statistical Mechanics and its Applications volume 324 ( year 2003 ) pages 133--140

    2003

  7. [15]

    Smith , author J

    author E. Smith , author J. D. Farmer , author L. s. Gillemot , author S. Krishnamurthy , et al., title Statistical theory of the continuous double auction , journal Quantitative Finance volume 3 ( year 2003 ) pages 481--514

    2003

  8. [16]

    Cont , author S

    author R. Cont , author S. Stoikov , author R. Talreja , title A stochastic model for order book dynamics , journal Operations Research volume 58 ( year 2010 ) pages 549--563

    2010

  9. [17]

    Abergel , author A

    author F. Abergel , author A. Jedidi , title A mathematical approach to order book modeling , journal International Journal of Theoretical and Applied Finance volume 16 ( year 2013 ) pages 1350025

    2013

  10. [18]

    Abergel , author A

    author F. Abergel , author A. Jedidi , title Long-time behavior of a hawkes process--based limit order book , journal SIAM Journal on Financial Mathematics volume 6 ( year 2015 ) pages 1026--1043

    2015

  11. [19]

    Paddrik , author R

    author M. Paddrik , author R. Hayes , author A. Todd , author S. Yang , author P. Beling , author W. Scherer , title An agent based model of the e-mini s&p 500 applied to flash crash analysis , journal 2012 IEEE Conference on Computational Intelligence for Financial Engineerin...

    2012

  12. [20]

    Vyetrenko , author D

    author S. Vyetrenko , author D. Byrd , author N. Petosa , author M. Mahfouz , author D. Dervovic , author M. Veloso , author T. Balch , title Get real: Realism metrics for robust limit order book market simulations , in: booktitle Proceedings of the First ACM International Con...

    2020

  13. [21]

    author R. L. Axtell , author J. D. Farmer , title Agent-based modeling in economics and finance: Past, present, and future , journal Journal of Economic Literature ( year 2022 ) pages 1--101

    2022

  14. [22]

    Coletta , et al., title Conditional generators for limit order book environments: Explainability, challenges, and robustness , journal arXiv preprint arXiv:2206.08351 ( year 2022 )

    author A. Coletta , et al., title Conditional generators for limit order book environments: Explainability, challenges, and robustness , journal arXiv preprint arXiv:2206.08351 ( year 2022 )

    2022 arXiv

  15. [23]

    Cont , author M

    author R. Cont , author M. Cucuringu , author J. Kochems , author F. Prenzel , title Limit order book simulation with generative adversarial networks , journal arXiv preprint arXiv:2306.12806 ( year 2023 )

    2023 arXiv

  16. [24]

    Hultin , et al., title A generative model of a limit order book using recurrent neural networks , journal Quantitative Finance volume 23 ( year 2023 ) pages 931--950

    author H. Hultin , et al., title A generative model of a limit order book using recurrent neural networks , journal Quantitative Finance volume 23 ( year 2023 ) pages 931--950

    2023

  17. [25]

    Cont , author M

    author R. Cont , author M. S. M \"u ller , title A stochastic partial differential equation model for limit order book dynamics , journal SIAM Journal on Financial Mathematics volume 12 ( year 2021 ) pages 744--787

    2021

  18. [26]

    Bodor , author L

    author H. Bodor , author L. Carlier , title Stylized facts and market microstructure: An in-depth exploration of german bond futures market , journal arXiv preprint arXiv:2401.10722 ( year 2024 )

    2024 arXiv

  19. [27]

    Wu , author M

    author P. Wu , author M. Rambaldi , author J.-F. Muzy , author E. Bacry , title Queue-reactive hawkes models for the order flow , journal arXiv preprint arXiv:1901.08938 ( year 2019 )

    1901 arXiv

  20. [28]

    Bouchaud , author J

    author J.-P. Bouchaud , author J. Bonart , author J. Donier , author M. Gould , title Trades, quotes and prices: financial markets under the microscope , publisher Cambridge University Press , year 2018

    2018

  21. [29]

    Zarinelli , author M

    author E. Zarinelli , author M. Treccani , author J. D. Farmer , author F. Lillo , title Beyond the square root: Evidence for logarithmic dependence of market impact on size and participation rate , journal Market Microstructure and Liquidity volume 1 ( year 2015 ) pages 1550004

    2015

  22. [30]

    Bacry , author A

    author E. Bacry , author A. Iuga , author M. Lasnier , author C.-A. Lehalle , title Market impacts and the life cycle of investors orders , journal Market Microstructure and Liquidity volume 1 ( year 2015 ) pages 1550009

    2015

  23. [31]

    Bucci , author M

    author F. Bucci , author M. Benzaquen , author F. Lillo , author J.-P. Bouchaud , title Crossover from linear to square-root market impact , journal Physical review letters volume 122 ( year 2019 ) pages 108302

    2019

  24. [32]

    Sato , author K

    author Y. Sato , author K. Kanazawa , title Does the square-root price impact law belong to the strict universal scalings?: quantitative support by a complete survey of the tokyo stock exchange market , journal arXiv preprint arXiv:2411.13965 ( year 2024 )

    2024

  25. [33]

    Gabaix , author P

    author X. Gabaix , author P. Gopikrishnan , author V. Plerou , author H. E. Stanley , title A theory of power-law distributions in financial market fluctuations , journal Nature volume 423 ( year 2003 ) pages 267--270

    2003

  26. [34]

    Zhang , author S

    author Z. Zhang , author S. Zohren , author S. Roberts , title Deeplob: Deep convolutional neural networks for limit order books , journal IEEE Transactions on Signal Processing volume 67 ( year 2019 ) pages 3001--3012

    2019

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.