Pith. sign in

REVIEW 4 major objections 7 minor 1 cited by

TLOB: A Novel Transformer Model with Dual Attention for Price Trend Prediction with Limit Order Book Data

T0 review · 4 major / 7 minor · reviewed 2026-08-08 · deepseek-v4-flash

Pith's one-line read A dual-attention transformer for limit order book data outperforms state-of-the-art price-trend models on every dataset and horizon tested, with average F1 gains of 3.7 on FI-2010, 1.3 on Tesla, 7.7 on Intel, and 1.1 on Bitcoin.

desk verdict A useful empirical breadth and an honest write-up, but the SoTA claim is currently riding on input-feature differences that the baselines did not get. read the letter →

arxiv 2502.15757 v3 pith:MDH34QE4 submitted 2025-02-12 q-fin.ST cs.AIcs.LGq-fin.TR

classification q-fin.STcs.AIcs.LGq-fin.TR
keywords pricetrendpredictionlimitorderbookdualattentiontransformerF1-scoremarketefficiencylabelingmethodmid-price
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 argues that a transformer that applies attention separately to time and to order-book features—TLOB—outperforms existing price-trend models on every dataset and horizon tested, with average F1 gains of 3.7 on the FI-2010 benchmark, 1.3 on Tesla, 7.7 on Intel, and 1.1 on Bitcoin. It also shows that a deliberately simple MLP architecture reaches similar or better results on short horizons, suggesting that complex architectures are not strictly necessary for this task. The paper introduces a labeling scheme that decouples the smoothing window from the prediction horizon, removing a bias it identifies in earlier labels, and reports that stock predictability declined between 2012 and 2015 for Intel, consistent with growing market efficiency. When the trend threshold is set to the average bid-ask spread, reported F1 scores deteriorate, which the paper reads as evidence that classification accuracy does not automatically translate into trading profitability.

What carries the argument

The machinery is dual self-attention inside each TLOB block: temporal attention computes attention across the sequence of LOB snapshots, spatial attention computes attention across the price-volume feature dimension, and an MLPLOB block (two fully connected layers with GeLU activation, applied once across features and once across time) stands in for the transformer feed-forward network. A bilinear normalization layer adapts to batch statistics to handle non-stationarity, sinusoidal positional embeddings preserve snapshot order, and the proposed labeling rule uses separate window length $k$ and horizon $h$ parameters to define the trend from smoothed future and past mid-prices, removing the horizon bias of earlier labels.

What would settle it

Retrain DeepLOB and BiNCTABL using exactly the augmented inputs TLOB and MLPLOB receive (104 handcrafted features on FI-2010, concatenated order messages on TSLA-INTC) and compare F1 on all horizons; the central SoTA claim would be weakened if the gaps fall below the reported 3.7, 1.3, 7.7, and 1.1 points.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that a dual-attention transformer—one attention pass over LOB snapshots for temporal dependencies and a second over LOB features for spatial dependencies, followed by an MLP mixing block and preceded by bilinear normalization—produces the highest F1 scores among the compared methods on all four datasets (FI-2010, Tesla, Intel, Bitcoin) across all four horizons. The same study finds that a plain MLP with feature-mixing and temporal-mixing layers often matches or beats more elaborate baselines, and that the gap between the two proposed models is small on short horizons but grows on long horizons in favor of TLOB. The paper additionally claims that predictability is not static: on Intel data, F1 falls by 6.68 points from 2012 to 2015, and that replacing the class-balancing threshold with a spread-based threshold lowers performance, highlighting the difference between predictive accuracy and practical profitability.

Load-bearing premise

The comparison assumes the baseline models were given the same input information as TLOB and MLPLOB, but the paper adds 104 handcrafted FI-2010 features and NASDAQ order messages only to its own models.

Editorial extensions

If this is right

  • Short-horizon price-trend prediction can be handled by a simple MLP; the transformer's advantage is concentrated at horizons 50 and 100.
  • The decoupled smoothing window and horizon in the new labeling method can be applied to any model, so existing baselines may improve without architectural changes.
  • F1 scores on older data overstate current predictability; the measured 6.68-point drop on Intel between 2012 and 2015 means recent data is needed for realistic evaluation.
  • Replacing the class-balancing threshold with the average spread lowers F1, so classification accuracy on balanced labels is not the same as profitability after transaction costs.

Reading between the lines

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

  • A controlled test that feeds the same handcrafted features and order messages to the baseline models would isolate how much of the reported gain comes from the architecture versus the extra input information; the paper does not run that test.
  • The new labeling method's benefit could be quantified by applying it to an existing baseline such as DeepLOB while keeping the architecture fixed.
  • The predictability-decline result rests on one stock at two dates; repeating it over many stocks and years would show whether the trend is a market-wide effect.
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.

Referee Report

4 major / 7 minor

Summary. The paper proposes two deep learning models for price trend prediction from limit order book (LOB) data: MLPLOB, an MLP with separate feature-mixing and temporal-mixing layers, and TLOB, a transformer with dual temporal and spatial attention, a bilinear normalization layer, and an MLP-based feed-forward block. The models are evaluated on FI-2010, a NASDAQ dataset (TSLA and INTC), and a Bitcoin dataset for horizons 10, 20, 50, and 100, using F1-score. The authors claim that TLOB outperforms state-of-the-art baselines on every dataset and horizon, and that MLPLOB also surpasses the baselines except for the longest horizons on some datasets. Additional contributions include a labeling method that decouples the smoothing window from the prediction horizon, an empirical claim that stock price predictability declined by 6.68 F1-score between 2012 and 2015 for Intel, and an exploration of setting the classification threshold to the average spread. The paper also promises to release code.

Significance. If the reported comparisons are correct, the paper would be a useful contribution to the LOB forecasting literature: it would demonstrate that a well-tuned MLP can compete with more complex architectures, and that dual attention provides gains on longer horizons. The code release and the ablation study are positive elements. However, the central state-of-the-art claim hinges on a baseline comparison that may not be input-matched, because the proposed models use 104 additional handcrafted features on FI-2010 and order-message enrichment on TSLA/INTC, while the baselines appear not to have received the same inputs. The margins in several key cells are smaller than the stated effect of these extra inputs, so the significance of the architectural contribution is not yet established beyond reasonable doubt.

major comments (4)
  1. [Section 6.4, Section 7.1, Table 3] The FI-2010 comparison is not input-matched. Section 6.4 says that 'the FI-2010 dataset also contains 104 handcrafted features derived from the LOB, we used them in both our models' and that this choice improved F1-score by approximately 1. Section 7.1 says the baseline numbers in Table 3 are 'extracted from the benchmark of Prata et al. [31] since the settings are equal for the FI-2010 dataset.' These two statements are in tension: if the settings were truly equal, the baselines would also have used the 104 features, but then the reported gain of approximately 1 from using them would be ill-defined. In any case, the reader cannot verify that the baselines received the same 144-dimensional input. The margins of TLOB over BiNCTABL at h=10 (+0.45) and h=100 (+0.71) are smaller than the stated feature-induced gain, so the SoTA claim on FI-2010 is not established. Please re-run the baselines on the identical input representation, or provide explicit evidence from [31] that the baselines used the same features.
  2. [Section 6.4, Tables 4-5] For TSLA and INTC, the paper states in Section 6.4 that the authors 'augmented the LOB snapshots by concatenating them with the corresponding orders' for their own models, yielding 'an approximate improvement of 1.5 in the F1-score.' DeepLOB and BiNCTABL were trained locally on these datasets, but the paper never specifies whether these baselines received the same order-message enrichment. The TSLA h=10 improvement of TLOB over BiNCTABL is +1.81, which is comparable to the stated augmentation gain, so the comparison may reflect the additional input information rather than the model architecture. The authors must specify the exact feature representation used by each baseline and, ideally, re-run the baselines with the same enrichment before claiming state-of-the-art performance on TSLA/INTC.
  3. [Section 7.4, Table 7] The abstract's claim that 'stock price predictability has declined over time, -6.68 in F1-score' is supported only by Table 7, which compares a single day of INTC from 2012 with a single day in 2015, using only TLOB, with no variance estimates or statistical test. A one-day comparison is far too fragile to support a general claim about market efficiency. Please expand the experiment to multiple days and models, or considerably soften the claim and move it out of the abstract.
  4. [Section 7.1, footnote 8] The footnote says 'if we had taken the results reported in the individual papers, MLPLOB and TLOB would have still outperformed all the other models,' but this is not demonstrated with numbers. Because the original papers differ in data splits, feature sets, and evaluation protocols, this assertion does not substitute for a re-computed comparison. Please either provide the corresponding table or remove the claim.
minor comments (7)
  1. [Section 6.2, footnote 4] The BTC dataset is described as sampled at 250 ms in Section 6.2, but footnote 4 says it is sampled every 100 ms; please reconcile these values.
  2. [Table 3] The DLA result at h=100 (52.2) is much lower than at the other horizons; please verify this value and add a note if it is due to a failed run.
  3. [Appendix B] The precision-recall curves are referenced as 'Fig. ??', indicating a missing cross-reference; please fix the figure numbering.
  4. [Section 7.2] The assertion that TLOB and MLPLOB converge in 'less than half the epochs required by BiNCTABL and DeepLOB' is not supported by any figure or training curve; please add evidence or soften the statement.
  5. [Table 10] The MLPLOB learning rate is listed as 0.003, which is surprising given the TLOB value of 0.0001; please check whether this is a typo.
  6. [Abstract and Introduction] The abstract states that the code is released at a GitHub URL, while the last paragraph of the Introduction says 'We commit to releasing the code publicly'; please unify these statements to avoid ambiguity about availability.
  7. [Section 6.4] The sentence 'TLOB and MLPLOB have a higher number of parameters compared to SoTA LOB-based models' is confusing because the next sentence says they are still smaller than standard deep learning models; consider rewording for clarity.

Circularity Check

0 steps flagged · score 2.0 of 10

No derivation reduces to its own inputs; the low score reflects a self-cited FI-2010 benchmark and a feature-matching caveat, not circular reasoning.

full rationale

The paper's central derivations are not circular. The labels are defined by explicit formulas (Eqs. 5-7) with independently chosen thresholds, and the TLOB/MLPLOB architectures are not defined in terms of the F1 scores they later predict. The TSLA/INTC and BTC comparisons are local re-trainings of the baseline models, so the main claim has independent grounding outside any fitted parameter or self-citation. The FI-2010 baseline numbers are taken from Prata et al. [31], a benchmark paper that includes one of the present authors; this is a self-citation, but the cited numbers are fixed external results and are not derived from the present paper's model assumptions. Section 6.4 states that the proposed models used the 104 FI-2010 handcrafted features and order-message augmentation for TSLA/INTC, while Section 7.1 says the FI-2010 baselines were extracted from [31] on the asserted basis of equal settings; this is a benchmarking-validity risk about input matching, not a circular reduction. No fitted parameter is renamed as a prediction, no equation is equivalent to its own input by construction, and no load-bearing uniqueness claim is imported from the authors' prior work. The predictability-decline and spread-threshold experiments are direct empirical comparisons on held-out data. The score of 2 reflects the minor self-citation and the input-matching caveat, not circularity in the derivation chain.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the trained model weights, which are learned from data, and on several hyperparameters chosen by hand. No new physical or conceptual entities are introduced. The key free parameter is the labeling threshold theta, set to balance classes rather than derived from a structural model. The axioms are standard domain assumptions in LOB forecasting, with the strongest unstated assumption being the comparability of the FI-2010 baseline numbers.

free parameters (5)
  • Label threshold theta for TSLA-INTC and BTC = mean percentage change, per dataset
    Chosen to balance class distribution (Section 6.4), not derived from trading costs; affects all reported F1 scores on these datasets.
  • Sequence size T = 128 for TLOB, 384 for MLPLOB
    Selected via grid search (Appendix A); a hyperparameter chosen by hand.
  • Learning rate = 0.0001 for TLOB, 0.003 for MLPLOB
    Selected via grid search; a hyperparameter chosen by hand.
  • Number of layers = 4 for TLOB, 3 for MLPLOB
    Selected via grid search; a hyperparameter chosen by hand.
  • Number of attention heads = 1
    Authors state that performance was equal for 1, 2, and 4 heads, so they fixed it to 1 (Appendix A).
assumptions (4)
  • domain assumption The FI-2010 benchmark labels and the baseline numbers from Prata et al. [31] are accurate and directly comparable.
    The paper uses these numbers without re-running the baselines under identical conditions (Section 7.1).
  • domain assumption Volume-based sampling of LOB data is a valid proxy for market activity and preserves predictive signal.
    Introduced in Section 6.3 without empirical justification that it is superior to time- or event-based sampling for the prediction task.
  • domain assumption Bilinear Normalization [37] is an effective normalization layer for non-stationary LOB data.
    Adopted from prior work without independent verification in this paper; cited as robust under distribution shift.
  • domain assumption Stylized facts and market microstructure characteristics are independent of individual stock behavior, so models trained on one stock transfer to another.
    Invoked in Section 6.3 to justify using the same model architecture and hyperparameters across TSLA and INTC.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TLOB: A Novel Transformer Model with Dual Attention for Price Trend Prediction with Limit Order Book Data." pith.science (2026). https://pith.science/paper/MDH34QE4

@misc{pith2026250215757,
  author       = {Pith},
  title        = {Pith review of: TLOB: A Novel Transformer Model with Dual Attention for Price Trend Prediction with Limit Order Book Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MDH34QE4}},
  note         = {Machine review of arXiv:2502.15757}
}
read the original abstract

Price Trend Prediction (PTP) based on Limit Order Book (LOB) data is a fundamental challenge in financial markets. Despite advances in deep learning, existing models fail to generalize across different market conditions and assets. Surprisingly, by adapting a simple MLP-based architecture to LOB, we show that we surpass SoTA performance; thus, challenging the necessity of complex architectures. Unlike past work that shows robustness issues, we propose TLOB, a transformer-based model that uses a dual attention mechanism to capture spatial and temporal dependencies in LOB data. This allows it to adaptively focus on the market microstructure, making it particularly effective for longer-horizon predictions and volatile market conditions. We also introduce a new labeling method that improves on previous ones, removing the horizon bias. We evaluate TLOB's effectiveness across four horizons, using the established FI-2010 benchmark, a NASDAQ and a Bitcoin dataset. TLOB outperforms SoTA methods in every dataset and horizon. Additionally, we empirically show how stock price predictability has declined over time, -6.68 in F1-score, highlighting the growing market efficiency. Predictability must be considered in relation to transaction costs, so we experimented with defining trends using an average spread, reflecting the primary transaction cost. The resulting performance deterioration underscores the complexity of translating trend classification into profitable trading strategies. We argue that our work provides new insights into the evolving landscape of stock price trend prediction and sets a strong foundation for future advancements in financial AI. We release the code at https://github.com/LeonardoBerti00/TLOB.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Exploring Microstructural Dynamics in Cryptocurrency Limit Order Books: Better Inputs Matter More Than Stacking Another Hidden Layer

    cs.LG 2025-06 reject novelty 4.0 of 10

    On one day of BTC/USDT order book data, Savitzky-Golay smoothing and feature choice helped simple models match or beat deeper neural networks.

Reference graph

Works this paper leans on

46 extracted references · 38 canonical work pages · cited by 1 Pith paper

  1. [31]

    Prata, G

    M. Prata, G. Masi, L. Berti, V . Arrigoni, A. Coletta, I. Cannistraci, S. Vyetrenko, P. Velardi, and N. Bartolini. Lob-based deep learning models for stock price trend prediction: a benchmark study. Artificial Intelligence Review, 57(5):1–45, 2024

  2. [1]

    Aiolfi and C

    M. Aiolfi and C. A. Favero. Model uncertainty, thick modelling and the predictability of stock returns. Journal of Forecasting, 24(4):233–254, 2005

  3. [2]

    Berti, B

    L. Berti, B. Prenkaj, and P. Velardi. Trades: Generating realistic market simulations with diffusion models. arXiv preprint arXiv:2502.07071, 2025

  4. [3]

    Bossaerts and P

    P. Bossaerts and P. Hillion. Implementing statistical criteria to select return forecasting models: what do we learn? The Review of Financial Studies, 12(2):405–428, 1999

  5. [4]

    Bouchaud, J

    J. Bouchaud, J. Bonart, J. Donier, and M. Gould. Trades, Quotes and Prices: Financial Markets Under the Microscope. Cambridge University Press, 2018. ISBN 9781107156050. URL https://books.google.it/books? id=u45LDwAAQBAJ

  6. [5]

    Bouchaud, M

    J.-P. Bouchaud, M. Mézard, and M. Potters. Statistical properties of stock order books: empirical results and models. Quantitative finance, 2 (4):251, 2002

  7. [6]

    Bouchaud, J

    J.-P. Bouchaud, J. D. Farmer, and F. Lillo. How markets slowly digest changes in supply and demand. In Handbook of financial markets: dynamics and evolution, pages 57–160. Elsevier, 2009

  8. [7]

    T. B. Brown. Language models are few-shot learners. arXiv preprint arXiv:2005.14165, 2020

Show all 46 references
  1. [8]

    D. Byrd, M. Hybinette, and T. H. Balch. Abides: Towards high-fidelity multi-agent market simulation. In Proceedings of the 2020 ACM SIGSIM Conference on Principles of Advanced Discrete Simulation, pages 11–22, 2020

  2. [9]

    X. Chen, C. Liang, D. Huang, E. Real, K. Wang, H. Pham, X. Dong, T. Luong, C.-J. Hsieh, Y . Lu, et al. Symbolic discovery of optimization algorithms. Advances in neural information processing systems , 36, 2024

  3. [10]

    Coletta, M

    A. Coletta, M. Prata, M. Conti, E. Mercanti, N. Bartolini, A. Moulin, S. Vyetrenko, and T. Balch. Towards realistic market simulations: a generative adversarial networks approach. In Proceedings of the Second ACM International Conference on AI in Finance, pages 1–9, 2021

  4. [11]

    Coletta, A

    A. Coletta, A. Moulin, S. Vyetrenko, and T. Balch. Learning to simulate realistic limit order book markets from data as a world agent. In Pro- ceedings of the Third ACM International Conference on AI in Finance, pages 428–436, 2022

  5. [12]

    R. Cont. Empirical properties of asset returns: stylized facts and statisti- cal issues. Quantitative finance, 1(2):223, 2001

  6. [13]

    R. Cont. Statistical modeling of high-frequency financial data. IEEE Signal Processing Magazine, 28(5):16–25, 2011

  7. [14]

    R. Cont, A. Kukanov, and S. Stoikov. The price impact of order book events. Journal of financial econometrics, 12(1):47–88, 2014

  8. [15]

    P. K. Diederik. Adam: A method for stochastic optimization. (No Title), 2014

  9. [16]

    Dimson and P

    E. Dimson and P. Marsh. Murphy’s law and market anomalies. Journal of Portfolio Management, 25(2):53–69, 1999

  10. [17]

    M. D. Gould, M. A. Porter, S. Williams, M. McDonald, D. J. Fenn, and S. D. Howison. Limit order books. Quantitative Finance, 13(11): 1709–1742, 2013

  11. [18]

    Guo and X

    Y . Guo and X. Chen. Forecasting the mid-price movements with high- frequency lob: a dual-stage temporal attention-based deep learning archi- tecture. Arabian Journal for Science and Engineering, 48(8):9597–9618, 2023

  12. [19]

    Hendrycks and K

    D. Hendrycks and K. Gimpel. Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415, 2016

  13. [20]

    Hultin, H

    H. Hultin, H. Hult, A. Proutiere, S. Samama, and A. Tarighati. A gen- erative model of a limit order book using recurrent neural networks. Quantitative Finance, pages 1–28, 2023

  14. [21]

    Kaplan, S

    J. Kaplan, S. McCandlish, T. Henighan, T. B. Brown, B. Chess, R. Child, S. Gray, A. Radford, J. Wu, and D. Amodei. Scaling laws for neural language models. arXiv preprint arXiv:2001.08361, 2020

  15. [22]

    S. Khan, M. Naseer, M. Hayat, S. W. Zamir, F. S. Khan, and M. Shah. Transformers in vision: A survey. ACM computing surveys (CSUR), 54 (10s):1–41, 2022

  16. [23]

    Kisiel and D

    D. Kisiel and D. Gorse. Axial-lob: High-frequency trading with axial attention. In 2022 IEEE Symposium Series on Computational Intelli- gence (SSCI), pages 1327–1333, 2022. doi: 10.1109/SSCI51031.2022. 10022284

  17. [24]

    Kisiel and D

    D. Kisiel and D. Gorse. Axial-lob: High-frequency trading with axial at- tention. In 2022 IEEE Symposium Series on Computational Intelligence (SSCI), pages 1327–1333. IEEE, 2022

  18. [25]

    R. Kissell. Algorithmic trading methods: Applications using advanced statistics, optimization, and machine learning techniques . Academic Press, 2020

  19. [26]

    J. Li, X. Wang, Y . Lin, A. Sinha, and M. Wellman. Generating realistic stock market order streams. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 727–734, 2020

  20. [27]

    S. Nagel. Machine learning in asset pricing , volume 1. Princeton University Press, 2021

  21. [28]

    P. Nagy, S. Frey, S. Sapora, K. Li, A. Calinescu, S. Zohren, and J. Foerster. Generative ai for end-to-end limit order book modelling: A token-level autoregressive generative model of message flow using a deep state space network. arXiv preprint arXiv:2309.00638, 2023

  22. [29]

    Ntakaris, M

    A. Ntakaris, M. Magris, J. Kanniainen, M. Gabbouj, and A. Iosi- fidis. Benchmark dataset for mid-price forecasting of limit order book data with machine learning methods. http://urn.fi/urn:nbn:fi: csc-kata20170601153214969115. N/A

  23. [30]

    Passalis, A

    N. Passalis, A. Tefas, J. Kanniainen, M. Gabbouj, and A. Iosifidis. Deep adaptive input normalization for time series forecasting. IEEE trans- actions on neural networks and learning systems , 31(9):3760–3765, 2019

  24. [32]

    Shi and J

    Z. Shi and J. Cartlidge. State dependent parallel neural hawkes process for limit order book event stream prediction and simulation. In Proceed- ings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 1607–1615, 2022

  25. [33]

    Sirignano and R

    J. Sirignano and R. Cont. Universal features of price formation in financial markets: perspectives from deep learning. In Machine learning and AI in finance, pages 5–15. Routledge, 2021

  26. [34]

    J. A. Sirignano. Deep learning for limit order books. Quantitative Finance, 19(4):549–570, 2019

  27. [35]

    I. O. Tolstikhin, N. Houlsby, A. Kolesnikov, L. Beyer, X. Zhai, T. Un- terthiner, J. Yung, A. Steiner, D. Keysers, J. Uszkoreit, et al. Mlp-mixer: An all-mlp architecture for vision. Advances in neural information processing systems, 34:24261–24272, 2021

  28. [36]

    D. T. Tran, A. Iosifidis, J. Kanniainen, and M. Gabbouj. Temporal attention-augmented bilinear network for financial time-series data anal- ysis. IEEE transactions on neural networks and learning systems, 30(5): 1407–1418, 2018

  29. [37]

    D. T. Tran, J. Kanniainen, M. Gabbouj, and A. Iosifidis. Data normal- ization for bilinear structures in high-frequency financial time-series. In 2020 25th International Conference on Pattern Recognition (ICPR), pages 7287–7292. IEEE, 2021

  30. [38]

    Tsantekidis, N

    A. Tsantekidis, N. Passalis, A. Tefas, J. Kanniainen, M. Gabbouj, and A. Iosifidis. Forecasting stock prices from the limit order book using con- volutional neural networks. In 2017 IEEE 19th conference on business informatics (CBI), volume 1, pages 7–12. IEEE, 2017

  31. [39]

    Tsantekidis, N

    A. Tsantekidis, N. Passalis, A. Tefas, J. Kanniainen, M. Gabbouj, and A. Iosifidis. Using deep learning to detect price change indications in financial markets. In 2017 25th European Signal Processing Conference (EUSIPCO), pages 2511–2515. IEEE, 2017

  32. [40]

    Tsantekidis, N

    A. Tsantekidis, N. Passalis, A. Tefas, J. Kanniainen, M. Gabbouj, and A. Iosifidis. Using deep learning for price prediction by exploiting stationary limit order book features. Applied Soft Computing, 93:106401, 2020

  33. [41]

    Vaswani, N

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017

  34. [42]

    Q. Wen, T. Zhou, C. Zhang, W. Chen, Z. Ma, J. Yan, and L. Sun. Trans- formers in time series: A survey. arXiv preprint arXiv:2202.07125 , 2022

  35. [43]

    A. Zeng, M. Chen, L. Zhang, and Q. Xu. Are transformers effective for time series forecasting? In Proceedings of the AAAI conference on artificial intelligence, volume 37, pages 11121–11128, 2023

  36. [44]

    Zhang and S

    Z. Zhang and S. Zohren. Multi-horizon forecasting for limit order books: Novel deep learning approaches and hardware acceleration using intelligent processing units, 2021. URL https://arxiv.org/abs/2105.10430

  37. [45]

    Zhang, S

    Z. Zhang, S. Zohren, and S. Roberts. Deeplob: Deep convolutional neural networks for limit order books. IEEE Transactions on Signal Processing, 67(11):3001–3012, 2019

  38. [46]

    Zhang, S

    Z. Zhang, S. Zohren, and S. Roberts. Deep reinforcement learning for trading. The Journal of Financial Data Science, 2(2):25–40, 2020. A Hyperparameters Search To find the best hyperparameters, we employ a grid search exploring different values as shown in Table 10. Regarding ...

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.