Pith. sign in

REVIEW 5 major objections 7 minor 42 references

NewsNet-SDF: Stochastic Discount Factor Estimation with Pretrained Language Model News Embeddings via Adversarial Networks

T0 review · 5 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read NewsNet-SDF claims a stochastic discount factor built from news embeddings and adversarial networks prices U.S. equities out-of-sample with a 2.80 Sharpe ratio.

desk verdict A credible architectural novelty undermined by an under-specified news pipeline and no code; the 2.80 Sharpe is a promise, not yet a result. read the letter →

arxiv 2505.06864 v1 pith:KTWBVIKF submitted 2025-05-11 q-fin.PM cs.LG

Shunyao Wang , Ming Cheng , Christina Dan Wang This is my paper · ORCID
classification q-fin.PMcs.LG MSC 91B2591G1068T50
keywords StochasticDiscountFactorPretrainedLanguageModelsAdversarialNetworksFinancialNewsEmbeddingsAssetPricingMultimodalAnalysisDeepLearningSDFEstimation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that unstructured financial news can be folded directly into a stochastic discount factor (the pricing kernel that converts uncertain payoffs into today's prices) using pretrained language-model embeddings and adversarial training. On U.S. stock data from 1980-2022, the resulting NewsNet-SDF model reports an out-of-sample Sharpe ratio of 2.80, beating CAPM by 471% and the five-factor model by 460%, with pricing errors 74% lower than the five-factor benchmark. The paper argues that news text contributes unique systematic risk information: removing news cuts the Sharpe ratio by 41%, while removing macroeconomic features cuts it by 31%. If correct, the result would mean that narrative information, not just numerical characteristics, belongs inside the pricing kernel rather than as a separate sentiment overlay.

What carries the argument

The load-bearing mechanism is the adversarial minimax objective $\min_\phi \max_\psi \frac{1}{N}\sum_j \| E[M_{t+1} R^e_{t+1,j} g_{t,j}] \|^2 + \lambda(\|\phi\|_2^2 + \|\psi\|_2^2)$, which converts the generalized-method-of-moments orthogonality condition $E[M_{t+1}R^e_{t+1}g_t]=0$ into a game: the SDF network minimizes pricing errors, while a conditional network generates instruments $g_t$ that challenge the SDF where it prices worst. Text enters through GTE-multilingual sentence embeddings aggregated by learned self-attention, projected with principal components, and concatenated with an LSTM macro state and ranked firm characteristics.

What would settle it

Re-run the test period using only news articles published strictly before the start of month $t$, and refit the PCA on training data alone; if the Sharpe ratio falls from 2.80 to near the no-news ablation value (1.65), the look-ahead-free alignment is doing the work. Alternatively, shuffle news dates across months by a random permutation; a surviving 2.80 Sharpe would indicate the text signal is not date-specific.

Watch

Extended reading notes

Core claim

The central claim is that a stochastic discount factor $M_{t+1} = 1 - \sum_i w_{t,i} R^e_{t+1,i}$ can be learned adversarially from fused features consisting of an LSTM-processed macroeconomic state, cross-sectionally ranked firm characteristics, and self-attention-aggregated, PCA-projected news embeddings from a multilingual transformer. In the out-of-sample period 2000-2022, the model achieves a Sharpe ratio of 2.80, an MSPE of 0.56, and a cross-sectional $R^2$ of 0.11, with $\beta$-decile portfolios monotonically increasing in returns ($R^2 > 0.95$). The paper further asserts that news-derived principal components rank among the most influential drivers of SDF dynamics, and that stocks most sensitive to news embeddings earned 14.2% annual excess returns after controlling for standard risk factors.

Load-bearing premise

The 2.80 Sharpe ratio rests on the assumption that the news text assigned to each firm-month contains no information from after that month's returns are realized—that the temporal alignment has no look-ahead.

Editorial extensions

If this is right

  • If the model's pricing kernel is right, news text functions as a systematic risk factor, not just an alpha signal, and should be included in asset-pricing tests.
  • The adversarial network supplies test assets adaptively, so pricing errors and anomalies are evaluated against the hardest-to-price conditions the data can express.
  • The reported cross-sectional relation implies that firms with news-embedding sensitivity earn roughly 14.2% higher annual excess returns even after controlling for standard factors.
  • The ablation pattern implies that adding news helps more than adding macroeconomic state, suggesting text is not redundant with standard characteristics.
  • The monotone decile pattern implies the estimated SDF preserves risk-return ordering out of sample, including through the 2008 crisis and COVID-19.

Reading between the lines

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

  • A testable extension is to apply the same adversarial instrument-generator mechanism to other unstructured data (analyst reports, regulatory filings, social media) or other pricing equations, which the paper leaves for future work.
  • The claimed 2-3 week lead time of news narratives suggests a real-time trading strategy could rebalance on news embeddings faster than monthly factor models; the paper does not test transaction costs or capacity.
  • An immediate robustness check is whether the 2.80 Sharpe survives the cost of translating the SDF weights into a tradeable long-short portfolio, since the paper reports the raw SDF portfolio's Sharpe.
  • If the lead-time result generalizes, text-based embeddings could serve as early-warning indicators for regime shifts in risk premia, which would be relevant to risk management systems.
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 / 7 minor

Summary. The paper proposes NewsNet-SDF, a stochastic discount factor estimator that fuses pretrained language model news embeddings, LSTM-encoded macroeconomic state, and cross-sectionally ranked firm characteristics, and trains an SDF through an adversarial moment-condition objective. On U.S. equity data from 1970 to 2022, with approximately 2.5 million New York Times articles, the authors report an out-of-sample Sharpe ratio of 2.80 for the 2000-2022 period, claiming large improvements over CAPM, Fama-French factor models, and neural SDF baselines. The paper also presents ablations, feature-importance analysis, beta-sorted decile portfolio tests, and a COVID-period analysis. The central empirical contribution is the claim that news text, processed through GTE-multilingual embeddings and attention aggregation, is the dominant source of the model's performance.

Significance. If the reported out-of-sample performance is valid, the paper would be a significant contribution to the growing literature on machine learning and text-based asset pricing: it offers a concrete way to incorporate unstructured news text into a theoretically grounded SDF framework, and the 2.80 Sharpe ratio is far above typical factor-model benchmarks. The architecture is coherent and follows the moment-based SDF logic of Hansen and Singleton, and the inclusion of three data modalities with an adversarial training objective is a sensible design. However, the significance is conditional on empirical credibility. The paper does not release code or data, does not report uncertainty around the headline Sharpe ratio, and leaves the news-to-return temporal alignment underspecified. Because the ablation attributes 41% of the Sharpe ratio to news text, these omissions are load-bearing for the main claim.

major comments (5)
  1. [Section 3.1 and Section 4.1] The temporal alignment between news articles and monthly return periods is not specified, which is critical because the headline result, a Sharpe ratio of 2.80 in Table 2, depends on the news features N_{t,i} containing no information realized after the start of month t+1. The paper says "careful temporal alignment" in Section 4.1 but does not state: how NYT publication timestamps are converted to monthly features; whether articles published during month t are included in N_{t,i} before the return R_{t+1} is realized; how multi-firm articles are split; how firm-months without news are handled; and whether the PCA projection and the attention parameters W, b, v in Eqs. (4)-(6) are estimated on the training sample (1970-1994) only or on the full sample. If any test-period information enters the news embeddings, the 2.80 Sharpe ratio and the 41% ablation effect are artifacts. Please provide an exact pipeline description, including the cutoff rule, and a robustness check that drops the most recent week of news before each month-end.
  2. [Section 4.4, Table 2] The headline performance metrics are reported as point estimates with no measures of uncertainty. A Sharpe ratio of 2.80 estimated over 276 monthly observations, from 2000 to 2022, needs a standard error or a confidence interval, for example from a block bootstrap or Newey-West correction. Without such intervals, the statement that NewsNet-SDF exceeds CAPM by 471% and FF5 by 460% is not statistically interpretable, since the difference could be within sampling noise, especially under time-series dependence and estimation error in the SDF weights. Please report confidence intervals for SR, EV, XS-R2, and MSPE, and for the differences relative to the baselines.
  3. [Section 4.2, Table 1] The baseline implementations are not documented. Table 1 lists model names and one-line descriptions, but the paper does not state how each baseline is estimated on the same sample, what hyperparameters are used, whether the same 56 firm characteristics are used for all models, or how IPCA factors, GAN-SDF architectures, and TF-IDF/BERT text features are constructed. Without this information, the relative improvements in Table 2 are not reproducible, and the 471% improvement over CAPM is not a well-defined empirical statement unless the CAPM implementation is described in detail. Please provide pseudo-code, hyperparameter settings, or a released code repository.
  4. [Section 4.4, Figures 3 and 5] There are inconsistencies between the stated test window and the figures. Table 2 and Section 4.1 define the test period as 2000-2022, but Figure 3 plots cumulative returns from 1970-2022, which includes the training and validation samples, making it impossible to visually isolate out-of-sample performance. Figure 5 is captioned "Cumulative Returns of Beta-Sorted Decile Portfolios (2000-2024)", with text in Section 4.6 also referring to 2000-2024, even though the data section states the sample ends in 2022. These discrepancies need to be resolved, and the figures should either be restricted to the out-of-sample window or clearly marked with training, validation, and test regions.
  5. [Section 4.6, Eqs. (14), (24), (25)] The beta-sorted decile monotonicity is an internal consistency check, not an independent validation of the model's risk-return predictions. Since M_{t+1} = 1 - sum_i w_{t,i} R^e_{t+1,i} in Eq. (14), the model-implied beta in Eq. (25) is a deterministic transformation of the same weights w_{t,i} that produce the model's predicted expected returns; sorting on predicted beta is therefore essentially sorting on the model's own forecast. The perfect monotonicity and R^2>0.95 in Section 4.6 are consequence of this construction and do not provide independent evidence that high-risk assets earn higher returns. Please validate the model on external test assets, or on characteristic-sorted portfolios formed independently of the model, and report whether the monotonicity survives controlling for standard factor exposures.
minor comments (7)
  1. [Section 4.3, Eq. (21)] The definition of XS-R2 in Eq. (21) appears to use the model predictions in the denominator, whereas a cross-sectional R-squared should be based on the variance of realized mean returns; please clarify the formula and its intended interpretation.
  2. [Algorithm 1, step 7] The algorithm computes M_{t+1} = 1 - sum_i w_{t,i} R^e_{t+1,i} over a mini-batch, but the SDF is defined over the full cross-section in Eq. (14); please specify how the batch-level SDF is formed and how the unbalanced panel is handled in the loss.
  3. [Eqs. (17)-(18)] The loss in Eq. (18) averages firm-level moments, while Algorithm 1 step 9 computes a loss over individual observations; please align the notation and explain the aggregation rule.
  4. [Section 4.7] The claims of "14.2% annual excess returns" and "18-32% reduction in pricing errors" are presented without standard errors, sample sizes, or a description of the statistical test; please add the necessary detail.
  5. [References] The Shapley value method in Eq. (26) is attributed to reference [5], a paper on predicting returns with text data; please cite the original Shapley value source and define the notation more precisely.
  6. [Table 1] The table entry for Elasticnet-SDF is attributed to "Kelly et al., 2019" in the table body, which appears inconsistent with the reference list entry [42] for Kozak, Nagel, and Santosh; please correct the attribution.
  7. [Figures 1 and 2] Figures 1 and 2 present very similar architecture diagrams; please consider merging them or clearly distinguishing the information each figure is intended to convey.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the headline out-of-sample Sharpe ratio and pricing-error comparisons are external benchmarks, and the beta-sorted decile test is a genuine out-of-sample prediction rather than a reduction to the training objective.

full rationale

The derivation chain is self-contained. The SDF is estimated on a training period (1970-1994), validated on 1995-1999, and evaluated out-of-sample on 2000-2022. The headline Sharpe ratio (2.80), MSPE (0.56), and ablations are external comparisons against CAPM, FF5, IPCA-SDF, GAN-SDF, etc., not quantities fitted by the model's loss. The Section 4.6 beta-sorted decile analysis does not constitute circularity: although beta is defined via the model's SDF (Eq. 25), the decile portfolios are formed over 2000-2024, outside the training window, so the monotone pattern is an out-of-sample test of the model-implied risk-return relation, not a restatement of the training objective. The paper's main weakness is an underspecified news pipeline: Section 4.1 promises 'careful temporal alignment' but does not specify timestamp-to-month matching, multi-firm article splitting, treatment of firm-months without news, or whether PCA/attention parameters are fit only on training data. This is a data-leakage and correctness risk, not a circularity. There are no load-bearing self-citations, and no step of the derivation reduces to its own inputs by construction.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The method rests on standard SDF existence and on an unverified news alignment pipeline. All reported performance numbers depend on unreported hyperparameters and on the temporal-cleanliness of the news embeddings. The ledger has no invented entities.

free parameters (3)
  • PCA dimension dN for news embeddings = not reported
    Section 3.1 projects attention-pooled sentence embeddings (Eq. 6) to R^{d_N}; d_N is chosen but not stated, and it is unclear if PCA is fit on train or full sample.
  • Regularization coefficient lambda = not reported
    Eq. (17)-(18) include lambda weight decay; the value is not stated.
  • Network hyperparameters (layer widths, attention dimension d_a, LSTM hidden size d_I, learning rates, batch size… = not reported
    Section 3 gives architecture forms but no numerical hyperparameters or training details, though these materially affect the reported SR.
assumptions (4)
  • domain assumption A stochastic discount factor exists and satisfies E[M_{t+1} R^e_{t+1,i}] = 0 under no-arbitrage.
    Section 2.1 Eq. 1; the entire loss function enforces a sample analog of this condition.
  • domain assumption GTE-multilingual-base sentence embeddings preserve financial semantics relevant to returns.
    Section 3.1; the model trusts a general-purpose pretrained encoder for finance text without domain fine-tuning or validation.
  • standard math Random sampling and adversarial training make sample moments consistent for population moments.
    Section 3.3 replaces population expectations in Eq. 17 with sample averages in Eq. 18; stationarity/ergodicity and saddle-point convergence are assumed but not discussed.
  • ad hoc to paper News articles can be assigned to a single firm and to the correct calendar period without look-ahead.
    Section 4.1 mentions 'careful temporal alignment' but provides no matching or missing-news rule; this assumption is load-bearing for the out-of-sample claim.

how reviews work

0 comments
Cite this review

Pith. "Pith review of NewsNet-SDF: Stochastic Discount Factor Estimation with Pretrained Language Model News Embeddings via Adversarial Networks." pith.science (2026). https://pith.science/paper/KTWBVIKF

@misc{pith2026250506864,
  author       = {Pith},
  title        = {Pith review of: NewsNet-SDF: Stochastic Discount Factor Estimation with Pretrained Language Model News Embeddings via Adversarial Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KTWBVIKF}},
  note         = {Machine review of arXiv:2505.06864}
}
read the original abstract

Stochastic Discount Factor (SDF) models provide a unified framework for asset pricing and risk assessment, yet traditional formulations struggle to incorporate unstructured textual information. We introduce NewsNet-SDF, a novel deep learning framework that seamlessly integrates pretrained language model embeddings with financial time series through adversarial networks. Our multimodal architecture processes financial news using GTE-multilingual models, extracts temporal patterns from macroeconomic data via LSTM networks, and normalizes firm characteristics, fusing these heterogeneous information sources through an innovative adversarial training mechanism. Our dataset encompasses approximately 2.5 million news articles and 10,000 unique securities, addressing the computational challenges of processing and aligning text data with financial time series. Empirical evaluations on U.S. equity data (1980-2022) demonstrate NewsNet-SDF substantially outperforms alternatives with a Sharpe ratio of 2.80. The model shows a 471% improvement over CAPM, over 200% improvement versus traditional SDF implementations, and a 74% reduction in pricing errors compared to the Fama-French five-factor model. In comprehensive comparisons, our deep learning approach consistently outperforms traditional, modern, and other neural asset pricing models across all key metrics. Ablation studies confirm that text embeddings contribute significantly more to model performance than macroeconomic features, with news-derived principal components ranking among the most influential determinants of SDF dynamics. These results validate the effectiveness of our multimodal deep learning approach in integrating unstructured text with traditional financial data for more accurate asset pricing, providing new insights for digital intelligent decision-making in financial technology.

Figures

Figures reproduced from arXiv: 2505.06864 by the authors.

Figure 1
Figure 1. NewsNet-SDF architecture integrating financial news, macroeconomic data, and firm characteristics through specialized pathways into a moment-based SDF frame￾work that enforces the fundamental pricing equation Et[Mt+1R e t+1] = 0. 2. We enhance SDF estimation through a novel loss function that dynamically reweights pricing errors based on factor exposure, reducing pricing errors by 74% compared to the Fama-French fiv… view at source ↗
Figure 2
Figure 2. NewsNet-SDF framework architecture: Specialized feature extraction pathways for news, macroeconomic data, and firm characteristics feed into an adversarial network structure that enforces the fundamental asset pricing equation E[Mt+1R e t+1] = 0. markets, and state-of-the-art performance on semantic textual similarity tasks, which is crucial for capturing nuanced market narratives. PLM-Based Embedding Generation. Fo… view at source ↗
Figure 3
Figure 3. Cumulative Returns of Different Models (1970-2022) of digital financial media, increasing the relevance of text-based signals; the emer￾gence of more complex market dynamics driven by algorithmic trading and social media; and post-crisis regulatory changes creating subtle market patterns that traditional factors fail to capture. Our approach excels precisely when informa￾tion complexity increases, extracting predict… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Top 50 Feature Importance Based on Average Sensitivity We quantify each feature’s influence using average sensitivity (Si), calculated as: Si = 1 n Xn j=1 [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]
Figure 5
Figure 5. Figure 5: Cumulative Returns of Beta-Sorted Decile Portfolios (2000-2024) [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: News embedding contribution analysis during market volatility. Panel A: quar￾terly importance scores by industry. Panel B: key semantic themes with high market impact. Following [5], we quantify embedding importance using the Shapley attribu￾tion method: ImportanceScor…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 37 canonical work pages

  1. [1]

    The Journal of Finance19(3), 425–442 (1964)

    Sharpe, W.F.: Capital asset prices: A theory of market equilibrium under condi- tions of risk. The Journal of Finance19(3), 425–442 (1964)

    work page 1964

  2. [2]

    Journal of Financial Economics33(1), 3–56 (1993)

    Fama, E.F., French, K.R.: Common risk factors in the returns on stocks and bonds. Journal of Financial Economics33(1), 3–56 (1993)

    work page 1993

  3. [3]

    The Review of Financial Studies33(5), 2223–2273 (2020)

    Gu, S., Kelly, B., Xiu, D.: Empirical asset pricing via machine learning. The Review of Financial Studies33(5), 2223–2273 (2020)

    work page 2020

  4. [4]

    Journal of Financial Economics134(3), 501–524 (2019)

    Kelly, B.T., Pruitt, S., Su, Y.: Characteristics are covariances: A unified model of risk and return. Journal of Financial Economics134(3), 501–524 (2019)

    work page 2019

  5. [5]

    NBER Working Paper No

    Ke, Z.T., Kelly, B.T., Xiu, D.: Predicting returns with text data. NBER Working Paper No. 26186 (2019)

    work page 2019

  6. [6]

    The Journal of Finance66(1), 35–65 (2011)

    Loughran, T., McDonald, B.: When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks. The Journal of Finance66(1), 35–65 (2011)

    work page 2011

  7. [7]

    The Journal of Finance62(3), 1139–1168 (2007)

    Tetlock, P.C.: Giving content to investor sentiment: The role of media in the stock market. The Journal of Finance62(3), 1139–1168 (2007)

    work page 2007

  8. [8]

    The Journal of Finance68(3), 1267–1300 (2013)

    Garcia, D.: Sentiment during recessions. The Journal of Finance68(3), 1267–1300 (2013)

    work page 2013

Show all 42 references
  1. [9]

    Princeton University Press (2009)

    Cochrane, J.H.: Asset Pricing: Revised Edition. Princeton University Press (2009)

    2009

  2. [10]

    Econometrica50(5), 1269–1286 (1982)

    Hansen, L.P., Singleton, K.J.: Generalized instrumental variables estimation of nonlinear rational expectations models. Econometrica50(5), 1269–1286 (1982)

    1982

  3. [11]

    Critical Finance Review10(1), 207–264 (2021)

    Chen, A.Y., Zimmermann, T.: Open source cross-sectional asset pricing. Critical Finance Review10(1), 207–264 (2021)

    2021

  4. [12]

    Baker,M.,Wurgler,J.:Investorsentimentinthestockmarket.JournalofEconomic Perspectives21(2), 129–151 (2007)

    2007

  5. [13]

    Journal of Accounting Research58(4), 1041–1074 (2020)

    Li, F.: Textual analysis in accounting and finance: A survey. Journal of Accounting Research58(4), 1041–1074 (2020)

    2020

  6. [14]

    Review of Financial Studies35(6), 3011–3059 (2022)

    Lopez-Lira, A.: Risk factors that matter: Textual analysis of risk disclosures for the cross-section of returns. Review of Financial Studies35(6), 3011–3059 (2022)

    2022

  7. [15]

    Artificial Intelligence Review50(1), 49–73 (2018)

    Xing, F.Z., Cambria, E., Welsch, R.E.: Natural language based financial forecast- ing: A survey. Artificial Intelligence Review50(1), 49–73 (2018)

    2018

  8. [16]

    Econometrica50(4), 1029–1054 (1982)

    Hansen, L.P.: Large sample properties of generalized method of moments estima- tors. Econometrica50(4), 1029–1054 (1982)

    1982

  9. [17]

    Journal of Economic Theory13(3), 341–360 (1976)

    Ross, S.A.: The arbitrage theory of capital asset pricing. Journal of Economic Theory13(3), 341–360 (1976)

    1976

  10. [18]

    Latent Variables in Socio-Economic Models, 365–383 (1977)

    Geweke, J.: The dynamic factor analysis of economic time series. Latent Variables in Socio-Economic Models, 365–383 (1977)

    1977

  11. [19]

    Stock,J.H.,Watson,M.W.:Dynamicfactormodels.OxfordHandbookofEconomic Forecasting1(1), 35–59 (2011)

    2011

  12. [20]

    The Journal of Fi- nance52(1), 57–82 (1997)

    Carhart, M.M.: On persistence in mutual fund performance. The Journal of Fi- nance52(1), 57–82 (1997)

    1997

  13. [21]

    Advances in Neural Information Processing Systems 25, 1097–1105 (2012)

    Krizhevsky, A., Sutskever, I., Hinton, G.E.: ImageNet classification with deep con- volutional neural networks. Advances in Neural Information Processing Systems 25, 1097–1105 (2012)

    2012

  14. [22]

    Inter- national Conference on Advances in Computing, Communications and Informatics, 1643–1647 (2017)

    Selvin, S., Vinayakumar, R., Gopalakrishnan, E.A., Menon, V.K., Soman, K.P.: Stock price prediction using LSTM, RNN and CNN-sliding window model. Inter- national Conference on Advances in Computing, Communications and Informatics, 1643–1647 (2017)

    2017

  15. [23]

    Nature323(6088), 533–536 (1986) NewsNet-SDF 19

    Rumelhart, D.E., Hinton, G.E., Williams, R.J.: Learning representations by back- propagating errors. Nature323(6088), 533–536 (1986) NewsNet-SDF 19

    1986

  16. [24]

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

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

    1997

  17. [25]

    International Joint Conference on Neural Networks, 1419–1426 (2017)

    Nelson,D.M.Q.,Pereira,A.C.M.,deOliveira,R.A.:Stockmarket’spricemovement prediction with LSTM neural networks. International Joint Conference on Neural Networks, 1419–1426 (2017)

    2017

  18. [26]

    Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2141–2149 (2017)

    Zhang, L., Aggarwal, C., Qi, G.J.: Stock price prediction via discovering multi- frequency trading patterns. Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2141–2149 (2017)

    2017

  19. [27]

    International Joint Conference on Artificial Intelligence, 2627–2633 (2017)

    Qin, Y., Song, D., Chen, H., Cheng, W., Jiang, G., Cottrell, G.: A dual-stage attention-based recurrent neural network for time series prediction. International Joint Conference on Artificial Intelligence, 2627–2633 (2017)

    2017

  20. [28]

    Proceedings of the 26th ACM SIGKDD Interna- tional Conference on Knowledge Discovery and Data Mining, 1748–1758 (2020)

    Ding, Y., Liu, W., Bian, J., Zhang, D., Liu, T.Y.: Investor-Imitator: A framework for trading knowledge extraction. Proceedings of the 26th ACM SIGKDD Interna- tional Conference on Knowledge Discovery and Data Mining, 1748–1758 (2020)

    2020

  21. [29]

    Management Science 70(2), 714–750 (2024)

    Chen, L., Pelger, M., Zhu, J.: Deep learning in asset pricing. Management Science 70(2), 714–750 (2024)

    2024

  22. [30]

    Pro- ceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics, 3369–3381 (2021)

    Sawhney, R., Agarwal, S., Wadhwa, A., Shah, R.R.: Deep attentive learning for stock movement prediction from social media text and company correlations. Pro- ceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics, 3369–3381 (2021)

    2021

  23. [31]

    In: Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics, 4171–4186 (2019)

    Devlin, J., Chang, M.W., Lee, K., Toutanova, K.: BERT: Pre-training of deep bidirectional transformers for language understanding. In: Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics, 4171–4186 (2019)

    2019

  24. [32]

    arXiv preprint arXiv:1907.11692 (2019)

    Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., Stoyanov, V.: RoBERTa: A robustly optimized BERT pretraining approach. arXiv preprint arXiv:1907.11692 (2019)

    2019 arXiv

  25. [33]

    arXiv preprint arXiv:2308.03281 (2023)

    Li, J., Tang, T., Zhao, W.X., Wen, Z., Liu, Y., Nie, J.Y., Wen, J.R.: Towards general text embeddings with multi-stage contrastive learning. arXiv preprint arXiv:2308.03281 (2023)

    2023 arXiv

  26. [34]

    arXiv preprint arXiv:1908.10063 (2019)

    Araci, D.: FinBERT: Financial sentiment analysis with pre-trained language mod- els. arXiv preprint arXiv:1908.10063 (2019)

    2019 arXiv

  27. [35]

    arXiv preprint arXiv:2006.08097 (2020)

    Yang, Y., Uy, M.C.S., Huang, A.: FinBERT: A pretrained language model for financial communications. arXiv preprint arXiv:2006.08097 (2020)

    2020 arXiv

  28. [36]

    In: Proceedings of the AAAI Confer- ence on Artificial Intelligence,36(10), 10999–11007 (2022)

    Liu, Z., Huang, D., Huang, K., Li, Z., Zhao, J.: FinBERT-MRC: Financial domain- specific language model for Chinese market. In: Proceedings of the AAAI Confer- ence on Artificial Intelligence,36(10), 10999–11007 (2022)

    2022

  29. [37]

    The Review of Financial Studies33(5), 2326–2377 (2020)

    Freyberger, J., Neuhierl, A., Weber, M.: Dissecting characteristics nonparametri- cally. The Review of Financial Studies33(5), 2326–2377 (2020)

    2020

  30. [38]

    monthly stock returns

    Green, J., Hand, J.R., Zhang, X.F.: The characteristics that provide independent information about average U.S. monthly stock returns. The Review of Financial Studies30(12), 4389–4436 (2017)

    2017

  31. [39]

    Journal of Business & Economic Statistics34(4), 574–589 (2016)

    McCracken, M.W., Ng, S.: FRED-MD: A monthly database for macroeconomic research. Journal of Business & Economic Statistics34(4), 574–589 (2016)

    2016

  32. [40]

    The Review of Financial Studies21(4), 1455–1508 (2008)

    Welch, I., Goyal, A.: A comprehensive look at the empirical performance of equity premium prediction. The Review of Financial Studies21(4), 1455–1508 (2008)

    2008

  33. [41]

    Journal of Financial Economics116(1), 1–22 (2015)

    Fama, E.F., French, K.R.: A five-factor asset pricing model. Journal of Financial Economics116(1), 1–22 (2015)

    2015

  34. [42]

    Journal of Financial Economics135(2), 271–292 (2020)

    Kozak, S., Nagel, S., Santosh, S.: Shrinking the cross-section. Journal of Financial Economics135(2), 271–292 (2020)

    2020

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.