Pith. sign in

REVIEW 3 major objections 5 minor 41 references

Is attention truly all we need? An empirical study of asset pricing in pretrained RNN sparse and global attention models

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read The paper claims that pretrained RNN models with global self-attention or sliding-window sparse attention, fitted with causal masks, deliver strong absolute returns and downside-risk hedging in long-only value-weighted backtests on 420 larg

desk verdict Useful causal-mask correction and systematic comparison, but the headline Sortino ratios are built on a survivorship-biased universe and lack statistical support; central claim needs major revision. read the letter →

arxiv 2508.19006 v2 pith:II3GRPXS submitted 2025-08-26 q-fin.PR cs.LGecon.EMq-fin.CP

Shanyan Lai This is my paper · ORCID
classification q-fin.PRcs.LGecon.EMq-fin.CP
keywords attentionmechanismsassetpricingrecurrentneuralnetworksfactormodelsself-attentionsparsecausalmaskSortinoratio
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 whether attention mechanisms borrowed from natural-language processing improve machine-learning asset pricing, and specifically whether two attention variants mounted on a pretrained RNN—global self-attention and sliding-window sparse attention—beat vanilla RNN, LSTM, GRU, and other attention benchmarks. It reports that in long-only value-weighted backtests with 50 basis-point transaction costs, the global self-attention model and the sliding-window sparse attention model achieve annualized Sortino ratios of 2.0 and 1.80 in the period containing COVID-19, outperforming buy-and-hold and most benchmarks on downside risk. The models also produce positive out-of-sample residual alphas across all periods, which the paper interprets as evidence of persistent mispricing rather than model misspecification. A distinctive technical claim is that causal masks prevent future-information leakage that prior attention-based asset-pricing models ignored.

What carries the argument

The central object is the pre-trained RNN attention model: an MLP autoencoder compresses 182 factor-portfolio inputs to 70% of their dimension and imputes missing values; a two-layer tanh RNN with 64 then 32 neurons encodes the time series; an attention layer—either global self-attention (scaled dot-product of query, key, and value projections) or sliding-window sparse attention (the same projections but only over the most recent w time steps)—reweights past RNN hidden states; and a causal mask forces j ≤ t in the attention scores. A linear layer maps the attention output to next-month excess return. The causal mask is the load-bearing modification: it makes the attention weights respect tim

What would settle it

Run the same backtest on a point-in-time universe that includes stocks delisted or with missing data during 2013–2022, using delisting returns and no survivorship filter; if the global self-attention and sparse-attention models no longer beat buy-and-hold on Sortino ratio during the COVID period, the central performance claim fails. A second check: remove the causal masks and retrain on identical inputs; if Sortino ratios do not deteriorate, the paper's distinct claim about future-information leakage is not what drives the results.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that attention mechanisms designed for long text—global self-attention from the Transformer and sliding-window sparse attention from Longformer—can be mounted on a two-layer RNN, pretrained with an MLP autoencoder on firm-characteristic-sorted portfolio factors, and produce tradable long-only signals for 420 large-cap US stocks. The argument is carried by a causal mask that restricts each time step's attention to past time steps, addressing look-ahead bias the paper attributes to earlier attention-based asset-pricing work. In the value-weighted backtest over January 2013 to December 2021, the period that includes COVID-19, the global self-attention

Load-bearing premise

The results are computed on 420 large-cap stocks selected because they had no missing values in the test period and few missing values earlier, which excludes stocks that were delisted or stopped trading; if those excluded stocks are part of the real opportunity set, the reported downside-risk performance may not be achievable.

Editorial extensions

If this is right

  • If the claims hold, long-only strategies built on causal-masked RNN attention signals are viable for large-cap US equities, delivering higher Sortino ratios and lower drawdowns than buy-and-hold in volatile markets.
  • Attention-based asset-pricing studies that omit causal masks should be treated with suspicion, because their reported performance may include future information.
  • Sliding-window sparse attention is the safer default for practitioners because its performance is less sensitive to market-cap weighting than global self-attention.
  • The persistent positive residual alphas imply that the 182-factor set does not fully price the cross-section, leaving room for factor discovery.
  • Value weighting tends to compress model differences and improve risk-adjusted returns, meaning model rankings depend on portfolio construction.

Reading between the lines

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

  • Editorial inference: the survivorship-biased sample—only stocks with no missing testing-period values—likely inflates the reported Sortino ratios; a point-in-time universe including delisted stocks would probably lower them, though the relative ranking of the two attention models might survive.
  • Editorial inference: the trend-following signal is defined by agreement between predicted and realized signs; used as described, that agreement criterion is only known ex post, so the backtest may embed a mild look-ahead even with causal attention masks; a tradable version should act only on predicted signs.
  • Editorial inference: the same architecture could be transferred to higher-frequency or international data, where sparse attention's linear cost in window size would be a practical advantage; a natural test is whether the 2.0 Sortino persists out of sample after including delistings.
  • Editorial inference: the paper's comparison to unmasked Transformer models suggests a direct ablation—masked versus unmasked attention on identical RNN inputs—would quantify how much of the reported performance is due to the causal mask alone.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes pre-trained RNN attention models — a global self-attention model and a sliding-window sparse attention model — for asset pricing on the top 420 large-cap US stocks. It compares them with additive attention, Luong's three attention variants, vanilla RNN, LSTM, and GRU, using 182 firm-characteristic-sorted portfolio factors. The models are evaluated by out-of-sample predictive accuracy (R2, MSE, alpha, Diebold-Mariano tests) and by equal- and value-weighted portfolio backtests over three periods: pre-COVID (1911), COVID (2112), and post-COVID (2212). The headline claim is that the global self-attention and sparse attention models achieve annualized Sortino ratios of 2.00 and 1.80, respectively, in the value-weighted COVID period (Table 5), and that the sparse attention model is more stable across market-cap sizes. The paper emphasizes the use of causal masks to avoid future-data leakage, which is a legitimate and under-appreciated point in the attention-based asset pricing literature.

Significance. If the empirical claims were robust, the paper would make a useful contribution by systematically comparing mainstream attention mechanisms in an asset pricing setting, by proposing simplified RNN-attention architectures that reduce overfitting risk, and by demonstrating that causal masking prevents a known look-ahead bias in attention models. The three-period design, including the COVID-19 crisis, is a strength, and the use of Diebold-Mariano tests for predictive accuracy is appropriate. However, the central backtesting claims — the Sortino ratios and the 'outperforms' statements — are not statistically supported and are based on a stock universe selected using full test-period survival. The missing specification of the sparse attention window size also undermines reproducibility. The causal-mask contribution is independent and appears sound, but the empirical asset pricing conclusions need substantial additional support before the headline claims can be accepted.

major comments (3)
  1. [Section 3, Tables 4–5] The stock universe is selected using information unavailable at the start of the out-of-sample period. The 420 stocks are required to have 'no missing value in the testing period and less than 50% of missing values during the training period,' with the testing period extending to December 2022. This removes any stock that was delisted or had data interruptions during 2013–2022, thereby excluding distressed and failed firms by construction. The value-weighted portfolio results in Table 5, including the headline annualized Sortino ratios of 2.00 (self att) and 1.80 (sparse att) in the 2112 period, are computed on this survivorship-biased universe. The stated 'too-big-to-fail' and 'going concern' motivation does not cure the look-ahead: a practitioner in 2013 cannot know which 420 stocks will have complete data through 2022. A point-in-time universe construction, where eligibility is based
  2. [Section 5.2, Tables 4–5] No confidence intervals or statistical significance tests are reported for the differences in Sharpe ratios, Sortino ratios, or annualized returns across models. The Diebold-Mariano tests in Section 5.1 (Table 3) assess differences in absolute prediction errors, not differences in portfolio performance metrics. Consequently, statements such as 'the self att model outperforms all alternative models' and 'the sparse att model performs more stably' rest on point estimates that could easily be within sampling variation, especially given only 83–120 monthly observations per period. The paper should provide bootstrap or HAC-based confidence intervals for the Sharpe and Sortino ratio differences, or otherwise quantify the uncertainty around the rankings. Without this, the central empirical claim is not statistically secured.
  3. [Section 4.3.4 and Section 5] The sliding-window sparse attention model is a proposed core contribution, but the attention window size w is never specified. Equation (39) defines the attention set S_t = {s | max(0, t−w) ≤ s ≤ t}, but no value is given in the model description, data section, or empirical results. This hyperparameter directly controls the trade-off between computational cost and long-term dependency capture and affects the reported performance. The manuscript also omits the L1 regularization coefficient λ, the early-stopping patience parameter, and the exact latent dimension of the MLP autoencoder (70% of 182 is 127.4, which is not an integer). These details are necessary for reproducibility and for assessing the sensitivity of the results to model choices.
minor comments (5)
  1. [Throughout] Numerous typos and inconsistent terminology distract the reader: 'annulized' for 'annualized', 'BHV' vs 'BHE' for buy-and-hold, 'roling' for 'rolling', 'prevelent' for 'prevalent', 'cancatenate' for 'concatenate', 'pertaining' for 'pre-training', and 'Univiersity' in the author affiliation. The paper would benefit from careful proofreading.
  2. [Section 4.4] The text refers to 'Equation (65)' for the linear output of the RNN attention models, but the equation is actually numbered (41). The equation numbering throughout Section 4 appears shifted; please verify all cross-references.
  3. [Section 4.1] The MLP autoencoder latent dimension is described as '70% of the original input.' With 182 input factors, this implies a non-integer number of latent units. Please state the exact integer dimension used and how it was determined.
  4. [Section 3, Table 1] The period labels '1911', '2112', and '2212' are not immediately intuitive; although the text explains them, a sentence in Table 1 or the caption would improve clarity. Also, 'Observation No.' for the in-sample period is 672, but it is unclear whether the validation window is included in this count.
  5. [Section 5, Eq. (79)] In the maximum drawdown formula, the notation Cmax(t) and C(t) is introduced but the definition of 'highest value during time t' is imprecise. Please clarify whether this is the running maximum of the cumulative return path.

Circularity Check

0 steps flagged · score 2.0 of 10

No load-bearing circularity: the OOS asset-pricing results are computed from fitted models on held-out data; the only self-citation is motivational and not used to define the reported predictions.

full rationale

The paper's derivation chain is not circular in the sense defined here. The key reported quantities—OOS R2, MSE, alpha, DM statistics, and the value-weighted Sortino ratios (e.g., 2.00 for self att and 1.80 for sparse att in Period 2112)—are computed from fitted RNN/attention models applied to out-of-sample months; they are not defined in terms of the models' inputs or training targets. The self-citation to Lai [1] is used as a motivation ('Building on the findings of Lai [1]'), but the current RNN/attention architecture does not derive its predictive content from that prior paper, and no parameter or evaluation metric is defined by reference to Lai [1]. The attention mechanisms and causal masks are adopted from external sources (Vaswani et al. [11]; Beltagy et al. [12]), not from a self-citation chain. The Section 3 universe filter requiring 'no missing value in the testing period' is a genuine look-ahead/survivorship concern that affects the economic interpretation of the headline Sortino ratios, but it does not make any reported result equal to a fitted input by construction: the portfolio metrics still depend on the estimated model outputs and are not mathematically forced by the universe definition. Similarly, comparing many architectures and then highlighting the best performers is model selection, not a circular derivation. Thus, while the paper has a minor self-citation and a notable data-validity issue, no circular step reaches the threshold of Eq. X = Eq. Y by construction or a fitted parameter being renamed as a prediction.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The paper's central quantitative claims rest on a handful of hand-chosen hyperparameters and modeling assumptions, most notably the unreported sparse attention window size, the 70% latent compression, and the 50bp static transaction cost. No new physical or economic entities are introduced.

free parameters (5)
  • MLP autoencoder latent dimension ratio = 70% of original factors
    Chosen by hand without a stated selection rule; affects all downstream factor representations and model performance.
  • Sparse attention window size w = not reported
    Directly controls the memory range of the sparse attention model; never specified for the experiments, yet the model's backtest performance depends on it.
  • L1 regularization coefficient lambda = not reported
    Equation (63) says lambda requires pre-setting; the actual value is not given, and it directly affects model fit and overfitting.
  • Early stopping patience = not reported
    Used in Algorithm 2; the patience value is not reported, affecting the final trained model.
  • Transaction cost = 50 basis points
    Chosen from Ma et al. [26] and applied statically; net portfolio returns are sensitive to this assumption.
assumptions (3)
  • domain assumption The sign of the predicted excess return can be used as a tradable long-only signal without market impact or liquidity constraints.
    Used in Section 5.2 to construct the backtest; ignores slippage, borrow constraints, and discrete trading.
  • domain assumption The residual OOS prediction error (alpha in Eq. 69) is an economic measure of abnormal return or arbitrage opportunity.
    The paper interprets positive average alpha as evidence of persistent mispricing (Section 5.1), which conflates model bias with a tradable anomaly.
  • ad hoc to paper The MLP autoencoder latent dimension ratio of 70% is a valid, non-test-set-informed choice.
    Section 4.1 states the neuron number is configured to 70% of the original input without justification or sensitivity analysis.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Is attention truly all we need? An empirical study of asset pricing in pretrained RNN sparse and global attention models." pith.science (2026). https://pith.science/paper/II3GRPXS

@misc{pith2026250819006,
  author       = {Pith},
  title        = {Pith review of: Is attention truly all we need? An empirical study of asset pricing in pretrained RNN sparse and global attention models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/II3GRPXS}},
  note         = {Machine review of arXiv:2508.19006}
}
read the original abstract

This study investigates the pre-trained RNN attention models with the mainstream attention mechanisms, such as additive attention, Luong's three attentions, global self-attention and sliding window sparse attention, for the empirical asset pricing research on the top 420 large-cap US stocks. This is the first paper on the large-scale state-of-the-art (SOTA) attention mechanisms applied in the asset pricing context. They overcome the limitations of the traditional machine learning-based asset pricing, such as mis-capturing the temporal dependency and short memory. Moreover, the enforced causal masks in the attention mechanisms address the future data leaking issue ignored by the more advanced attention-based models, such as the classic Transformer. The proposed attention models also consider the temporal sparsity characteristic of asset pricing data and mitigate potential overfitting issues by deploying the simplified model structures. This provides some insights for future empirical economic research. All models are examined in three periods, which cover pre-COVID-19, COVID-19 and one year post-COVID-19, for testing the stability of these models under extreme market conditions. The study finds that in value-weighted portfolio back testing, the global self-attention model and the sliding window sparse attention model exhibit excellent capabilities in deriving the absolute returns and hedging downside risks, while they achieve an annualized Sortino ratio of 2.0 and 1.80 respectively in the period with COVID-19 in the static transaction cost scenario. Moreover, the sliding window sparse attention model performs more stably than the global self-attention model from the perspective of absolute portfolio returns with respect to the size of stocks' market capitalization.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 29 canonical work pages

  1. [1]

    Multilayer Perceptron Neural Network Models in Asset Pricing: An Empirical Study on Large-Cap US Stocks

    Lai, S.: Multilayer perceptron neural network models in asset pricing: An empirical study on large-cap us stocks. arXiv preprint arXiv:2505.01921 (2025)

    work page Pith review arXiv 2025

  2. [3]

    Review of Financial Studies 33(5), 2223–2273 (2020) https://doi.org/10.1093/ rfs/hhaa009

    Gu, S., Kelly, B., Xiu, D.: Empirical Asset Pricing via Machine Learning. Review of Financial Studies 33(5), 2223–2273 (2020) https://doi.org/10.1093/ rfs/hhaa009

    work page 2020

  3. [4]

    CRC Press, Boca Raton, FL (2020)

    Coqueret, G., Guida, T.: Machine Learning for Factor Investing: R Version. CRC Press, Boca Raton, FL (2020)

    work page 2020

  4. [5]

    Centre for Financial Research (CFR), working paper (2020)

    Andrew Y.; Zimmermann, T.C.: Open source cross-sectional asset pricing. Centre for Financial Research (CFR), working paper (2020)

    work page 2020

  5. [6]

    Available at SSRN 4999011 (2024)

    Zhou, X., Wang, Y.: Learning from memory: Asset pricing via recurrent neural network and attention mechanism. Available at SSRN 4999011 (2024)

    work page 2024

  6. [7]

    Cognitive science 14(2), 179–211 (1990)

    Elman, J.L.: Finding structure in time. Cognitive science 14(2), 179–211 (1990)

    work page 1990

  7. [8]

    nature 323(6088), 533–536 (1986)

    Rumelhart, D.E., Hinton, G.E., Williams, R.J.: Learning representations by back- propagating errors. nature 323(6088), 533–536 (1986)

    work page 1986

  8. [9]

    Neural computation 9(8), 1735–1780 (1997)

    Hochreiter, S., Schmidhuber, J.: Long short-term memory. Neural computation 9(8), 1735–1780 (1997)

    1997

Show all 41 references
  1. [10]

    In: EMNLP 2014 - 2014 Conference on Empirical Methods in Natural Language Processing, Proceedings of the Conference, pp

    Cho, K., Merri¨ enboer, B.V., Gulcehre, C., Bahdanau, D., Bougares, F., Schwenk, H., Bengio, Y.: Learning phrase representations using RNN encoder-decoder for statistical machine translation. In: EMNLP 2014 - 2014 Conference on Empirical Methods in Natural Language Processing,...

    2014 doi

  2. [11]

    Advances in neural information processing systems 30 (2017)

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

    2017

  3. [12]

    arXiv preprint arXiv:2004.05150 (2020)

    Beltagy, I., Peters, M.E., Cohan, A.: Longformer: The long-document trans- former. arXiv preprint arXiv:2004.05150 (2020)

    2004 arXiv

  4. [13]

    Technical report, National Bureau of Economic Research (2025)

    Kelly, B.T., Kuznetsov, B., Malamud, S., Xu, T.A.: Artificial intelligence asset pricing models. Technical report, National Bureau of Economic Research (2025)

    2025

  5. [14]

    Journal of Financial Economics 33(1), 3–56 (1993) https://doi.org/10

    Fama, E.F., French, K.R.: Common risk factors in the returns on stocks and bonds. Journal of Financial Economics 33(1), 3–56 (1993) https://doi.org/10. 47 1016/0304-405X(93)90023-5

    1993

  6. [15]

    Journal of Finance 52(1), 57–82 (1997) https://doi.org/10.1111/j.1540-6261.1997.tb03808.x

    Carhart, M.M.: On persistence in mutual fund performance. Journal of Finance 52(1), 57–82 (1997) https://doi.org/10.1111/j.1540-6261.1997.tb03808.x

    1997

  7. [16]

    Journal of Financial Economics 116(1), 1–22 (2015) https://doi.org/10.1016/j.jfineco.2014.10.010

    Fama, E.F., French, K.R.: A five-factor asset pricing model. Journal of Financial Economics 116(1), 1–22 (2015) https://doi.org/10.1016/j.jfineco.2014.10.010

    2015 doi

  8. [17]

    Review of Financial Studies 28(3), 650–705 (2015) https://doi.org/10.1093/rfs/ hhu068

    Hou, K., Xue, C., Zhang, L.: Digesting anomalies: An investment approach. Review of Financial Studies 28(3), 650–705 (2015) https://doi.org/10.1093/rfs/ hhu068

    2015 doi

  9. [18]

    Available at SSRN 3341728 (2019)

    Harvey, C.R., Liu, Y.: A census of the factor zoo. Available at SSRN 3341728 (2019)

    2019

  10. [19]

    International Journal of Data Science and Analytics 12(2), 175–183 (2021) https://doi.org/10

    Wang, Q.: Cryptocurrencies asset pricing via machine learning. International Journal of Data Science and Analytics 12(2), 175–183 (2021) https://doi.org/10. 1007/s41060-021-00252-6

    2021

  11. [20]

    Journal of Econometrics 222(1), 429–450 (2021)

    Gu, S., Kelly, B., Xiu, D.: Autoencoder asset pricing models. Journal of Econometrics 222(1), 429–450 (2021)

    2021

  12. [21]

    Available at SSRN 3450322 (2021)

    Avramov, D., Cheng, S., Metzker, L.: Machine learning versus economic restric- tions: Evidence from stock return predictability. Available at SSRN 3450322 (2021)

    2021

  13. [22]

    Management Science 70(2), 714–750 (2024) https://doi.org/10.1287/mnsc.2023.4695

    Chen, L., Pelger, M., Zhu, J.: Deep Learning in Asset Pricing. Management Science 70(2), 714–750 (2024) https://doi.org/10.1287/mnsc.2023.4695

    2024

  14. [23]

    Journal of Financial Economics 134(3), 501–524 (2019) https: //doi.org/10.1016/j.jfineco.2019.05.001

    Kelly, B.T., Pruitt, S., Su, Y.: Characteristics are covariances: A unified model of risk and return. Journal of Financial Economics 134(3), 501–524 (2019) https: //doi.org/10.1016/j.jfineco.2019.05.001

    2019 doi

  15. [24]

    Journal of Economic Surveys (2022) https://doi.org/10.1111/joes.12532

    Bagnara, M.: Asset Pricing and Machine Learning: A critical review. Journal of Economic Surveys (2022) https://doi.org/10.1111/joes.12532

    2022 doi

  16. [25]

    Advances in neural information processing systems 27 (2014)

    Goodfellow, I.J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., Bengio, Y.: Generative adversarial nets. Advances in neural information processing systems 27 (2014)

    2014

  17. [26]

    International Review of Financial Analysis 90 (2023) https://doi.org/10.1016/j.irfa.2023.102876

    Ma, T., Wang, W., Chen, Y.: Attention is all you need: An interpretable transformer-based asset allocation approach. International Review of Financial Analysis 90 (2023) https://doi.org/10.1016/j.irfa.2023.102876

    2023

  18. [27]

    arXiv preprint arXiv:2209.12014 (2022) 48

    Zhang, C.: Asset pricing and deep learning. arXiv preprint arXiv:2209.12014 (2022) 48

    2022 arXiv

  19. [28]

    arXiv preprint arXiv:2108.08999 (2021)

    Cong, L.W., Tang, K., Wang, J., Zhang, Y.: Deep sequence modeling: Devel- opment and applications in asset pricing. arXiv preprint arXiv:2108.08999 (2021)

    2021 arXiv

  20. [29]

    Available at SSRN 3971876 (2021)

    Chatigny, P., Goyenko, R., Zhang, C.: Asset pricing with attention guided deep learning. Available at SSRN 3971876 (2021)

    2021

  21. [30]

    In: 3rd International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings, pp

    Bahdanau, D., Cho, K.H., Bengio, Y.: Neural machine translation by jointly learning to align and translate. In: 3rd International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings, pp. 1–15 (2015)

    2015

  22. [31]

    arXiv preprint arXiv:1508.04025 (2015)

    Luong, M.-T., Pham, H., Manning, C.D.: Effective approaches to attention-based neural machine translation. arXiv preprint arXiv:1508.04025 (2015)

    2015 arXiv

  23. [32]

    In: Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp

    Akhtar, M.S., Kumar, A., Ghosal, D., Ekbal, A., Bhattacharyya, P.: A multilayer perceptron based ensemble technique for fine-grained financial sentiment anal- ysis. In: Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pp. 540–546 (2017)

    2017

  24. [33]

    Indonesian Journal of Electrical Engineering and Computer Science 24(3), 1759–1771 (2021)

    Sagiraju, H.K., Mogalla, S.: Application of multilayer perceptron to deep rein- forcement learning for stock market trading and analysis. Indonesian Journal of Electrical Engineering and Computer Science 24(3), 1759–1771 (2021)

    2021

  25. [34]

    PloS one12(7), 0180944 (2017)

    Bao, W., Yue, J., Rao, Y.: A deep learning framework for financial time series using stacked autoencoders and long-short term memory. PloS one12(7), 0180944 (2017)

    2017

  26. [35]

    arXiv preprint arXiv:2404.01866 (2024)

    Bieganowski, B., Slepaczuk, R.: Supervised autoencoder mlp for financial time series forecasting. arXiv preprint arXiv:2404.01866 (2024)

    2024 arXiv

  27. [36]

    Applied Stochastic Models in Business and Industry 33(1), 3–12 (2017)

    Heaton, J.B., Polson, N.G., Witte, J.H.: Deep learning for finance: deep portfolios. Applied Stochastic Models in Business and Industry 33(1), 3–12 (2017)

    2017

  28. [37]

    IEEE transactions on neural networks 5(2), 157–166 (1994)

    Bengio, Y., Simard, P., Frasconi, P.: Learning long-term dependencies with gra- dient descent is difficult. IEEE transactions on neural networks 5(2), 157–166 (1994)

    1994

  29. [38]

    Neural computation 12(10), 2451–2471 (2000)

    Gers, F.A., Schmidhuber, J., Cummins, F.: Learning to forget: Continual prediction with lstm. Neural computation 12(10), 2451–2471 (2000)

    2000

  30. [39]

    In: International Conference on Machine Learning, pp

    Jozefowicz, R., Zaremba, W., Sutskever, I.: An empirical exploration of recurrent network architectures. In: International Conference on Machine Learning, pp. 2342–2350 (2015). PMLR

    2015

  31. [40]

    In: International Conference on Frontiers of Intelligent Computing: Theory and Applications, pp

    Galphade, M., Nikam, V., Banerjee, B., Kiwelekar, A.W.: Comparative analysis of wind power forecasting using lstm, bilstm, and gru. In: International Conference on Frontiers of Intelligent Computing: Theory and Applications, pp. 483–493 49 (2022). Springer

    2022

  32. [41]

    International Scientific and Practical Conference in Control Engineering and Decision Making, pp

    Pudikov, A., Brovko, A.: Comparison of lstm and gru recurrent neural net- work architectures. International Scientific and Practical Conference in Control Engineering and Decision Making, pp. 114–124 (2020). Springer

    2020

  33. [42]

    Diebold, F.X., Mariano, R.S.: Comparing predictive accuracy. Journal of Business and Economic Statistics 13(3), 253–263 (1995) 50 Appendix A Out-of-sample α distributions for alternative models (a) Batt(1911) (b) GRU(1911) (c) LSTM(1911) (d) Batt(2112) (e) GRU(2112) (f) LSTM(2...

    1995

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.