A Study of Dynamic Stock Relationship Modeling and S&P500 Price Forecasting Based on Differential Graph Transformer

Linyue Hu; Qi Wang

REVIEW 4 major objections 4 minor 1 cited by

A Study of Dynamic Stock Relationship Modeling and S&P500 Price Forecasting Based on Differential Graph Transformer

T0 review · 4 major / 4 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 This paper claims that modeling stock correlations as a changing graph, learned through a differential attention mechanism inside a Transformer, predicts next-day S&P 500 prices with about 90% lower RMSE than a GRU baseline.

desk verdict Plausible correlation-metric comparison undermined by two incompatible model definitions and a likely lookahead leak, so the headline numbers are unattributable. read the letter →

arxiv 2506.18717 v1 pith:JHVDSOYJ submitted 2025-06-23 cs.CE cs.AI

Linyue Hu , Qi Wang This is my paper · ORCID

classification cs.CEcs.AI

keywords DifferentialGraphTransformerdynamicstockrelationshipmodelingpricepredictionKendall'sTaucorrelationmulti-headattentionS&P500forecastinginterpretabilityanalysis

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 claims that stock price prediction improves when the model treats the web of stock correlations as something that changes over time, rather than as a fixed graph. It builds a Differential Graph Transformer (DGT) that combines causal temporal attention with differential graph attention, in which a learned difference between two attention maps keeps informative stock-to-stock edges and suppresses noise. Using ten years of z-score-normalized S&P 500 closing prices, the paper reports that DGT with global Kendall's Tau spatial priors achieves RMSE 0.238 and MAE 0.106, roughly 90% and 78% below the GRU baseline. It also reports that global correlation scopes generally beat local or hybrid scopes, and that K-means clustering separates high-volatility growth stocks from defensive blue chips, with the latter predicted far more accurately.

What carries the argument

The load-bearing mechanism is the differential graph attention head. For each head at time $t$, the model forms two softmax attention maps from separate query and key projections, multiplies the first elementwise by a predefined correlation adjacency matrix $A_t(h)$, subtracts the second scaled by a learnable $\lambda$, and multiplies by values: $\mathrm{head}_t(h) = (\mathrm{softmax}(Q_{t1}(h)K_{t1}(h)^\top/\sqrt{d/H}) \odot A_t(h) - \lambda\,\mathrm{softmax}(Q_{t2}(h)K_{t2}(h)^\top/\sqrt{d/H}))V_t(h)$. This lets the network retain strong edges from the prior graph while the subtracted term cancels redundant or noisy attention, so no threshold or top-$k$ sparsification is needed. The temporal branch uses causal masking so each day attends only to past days, and the spatial branch consumes correlation matrices built from four metrics (Pearson, mutual information, Spearman, Kendall's Tau) over three scopes (global, local 64-day blocks, and dual). Together the two branches are what the paper claims makes the model simultaneously capture time evolution and changing cross-stock structure.

What would settle it

Recompute Table 3 with every local and dual correlation matrix built only from trailing data up to the prediction day, using causal 64-day windows; if the RMSE and MAE improvements over GRU shrink or disappear, the reported 90% and 78% gains came from future information.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that a Transformer which injects a correlation-derived graph into a differential attention mechanism, and masks temporal attention causally, can forecast next-day S&P 500 constituent prices far more accurately than a recurrent baseline. The best configuration uses Kendall's Tau correlation matrices computed over the whole training period: RMSE 0.238 and MAE 0.106 on normalized prices, versus GRU RMSE 2.360 and MAE 0.478, which the paper describes as about 90% lower RMSE and 78% lower MAE. The same architecture without spatial information still beats GRU (RMSE 1.485), and adding spatial information cuts RMSE by about 41%, so the graph prior, not just the Transformer, is doing the work. The paper further reports that global correlation consistently outperforms local and dual scopes, and that high-growth, high-volatility stocks are harder to predict than defensive blue-chip stocks (Kendall global RMSE 0.481 versus 0.130).

Load-bearing premise

The load-bearing premise is that each 64-day local correlation matrix used when predicting day $t+1$ is computed without any prices after day $t$, even though the paper only describes it as a per-quarter matrix and never states that the calculation is causal.

Editorial extensions

If this is right

On z-score-normalized S&P 500 prices, DGT with global Kendall's Tau predicts next-day values with RMSE 0.238 and MAE 0.106, versus GRU RMSE 2.360 and MAE 0.478.
Adding spatial correlation information alone cuts DGT RMSE from 1.485 to 0.873, about 41%, before any correlation metric is chosen.
Global correlation scopes beat local and dual scopes for most metrics, with Kendall's Tau global giving the best overall numbers.
K-means clustering separates high-growth, high-volatility stocks from defensive blue chips; the defensive group is predicted much more accurately (Kendall global RMSE 0.130 versus 0.481).
Nonlinear correlation measures, especially Kendall's Tau and mutual information, outperform Pearson for high-volatility stocks.

Reading between the lines

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

If the causal local-correlation gap is closed and the results persist, the method is a practically usable non-lookahead forecaster; the reported numbers currently depend on an unstated causal assumption.
The consistent superiority of Kendall's Tau suggests that ordinal co-movement, not price magnitude, is the informative cross-stock signal, pointing toward rank-based portfolio strategies as a natural extension.
Because global correlations beat local ones, the model may be capturing stable market-regime structure rather than short-lived relationships; testing DGT on other markets or on crisis windows would show whether that global dominance is general.
The differential attention pattern of subtracting one softmax map from another is a generic sparse-attention prior that could be carried over to other noisy relational time series, such as supply-chain or macroeconomic networks.

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.

Desk Editor's Note

read the letter

Colleague,

This is a stock prediction paper that grafts a differential attention mechanism onto graph-structure priors and compares four correlation metrics across global/local/dual scopes on S&P 500 data. The systematic correlation-metric comparison and the cluster-level error analysis are the genuinely new parts. The finding that Kendall's Tau and global scope work best is plausible, and the relative ordering in Table 3 is internally consistent. The k-means split into high-volatility growth versus defensive blue-chip stocks, and the observation that the model does better on the latter, is a nice interpretability exercise. The paper also honestly lists limitations: price-only inputs, single market, simplified clustering.

The soft spots are serious. The differential graph attention mechanism is defined two incompatible ways. Section 2.3 adds an adjacency difference ΔA(t) to the softmax logits; Section 3.4 uses a two-stream softmax with the adjacency matrix multiplied elementwise, then a λ-weighted subtraction of the two streams. These are not equivalent and the paper never reconciles them. So we don't know which architecture actually produced Table 3's results. There's also a likely lookahead issue with local correlations: the 64-day block correlation is never stated to be causal, so if the full block is used to construct the graph for each time step inside it, future prices leak into the prediction. The abstract misreports the baseline: it says 'RMSE: 0.24 vs. 0.87' against GRU, but GRU's RMSE in Table 3 is 2.36; 0.87 is DGT with spatial information but no correlation matrix. That's not a typo, it's a substantive misstatement. On top of that, no code, no error bars, no seed variability, and a single GRU baseline.

Net: the empirical comparison may be salvageable if the model is specified precisely and the local correlation leakage is fixed, but as written the central claim is not attributable to a well-defined model. This is a desk reject for me. If the authors clean up the definition and release code, the correlation-metric comparison could be worth a look. Not something I'd cite or bring to the group as a positive example, though it's useful as a cautionary case.

Referee Report

4 major / 4 minor

Summary. The paper proposes a Differential Graph Transformer (DGT) for next-day S&P 500 constituent price prediction. The model combines causal temporal attention with spatial graph priors derived from four correlation measures (Pearson, mutual information, Spearman, Kendall's Tau) at global, local, and dual scopes. In a backtest on 470 stocks over 2015-2025, the authors report that DGT with global Kendall's Tau achieves RMSE 0.238 and MAE 0.106, about 90% and 78% lower than a GRU baseline, and that K-means clustering separates stocks into high-growth/high-volatility and defensive groups with different prediction errors.

Significance. The empirical scope is substantial: the paper systematically compares four correlation metrics across three temporal scopes and provides cluster-level analysis with statistical tests, which, if the model is well-specified and the results are reproducible, would be a useful reference for graph-based financial forecasting. The strength of the manuscript lies in the breadth of the correlation-metric comparison and the interpretable clustering of predictability differences. However, the current value is conditional because the architecture is not uniquely defined, the local-correlation protocol may leak future information, and the reported single-run results lack uncertainty quantification.

major comments (4)

[Sections 2.3 and 3.4] Sections 2.3 and 3.4 define two different differential attention mechanisms. Section 2.3 gives DiffAttention(Q,K,V)=softmax(QK^T/sqrt(d_k)+λΔA(t))V with ΔA(t)=A(t)-A(t-1), an additive logit adjustment from adjacency differences. Section 3.4 instead computes head_t(h)=(softmax(Q_t1 K_t1^T/sqrt(d/H))⊙A_t(h)-λsoftmax(Q_t2 K_t2^T/sqrt(d/H)))V_t(h), where the graph prior enters by multiplication and the 'difference' is between two softmax attention maps. These are not algebraically or conceptually equivalent, and the text at the start of Section 3.4 only says the latter 'continues the basic principle' of Section 2.3. Since no code is released, Table 3, including the headline global Kendall's Tau row (RMSE 0.238, MAE 0.106), cannot be attributed to a uniquely specified architecture. Please state which formula was implemented, reconcile the two definitions, and release code.
[Section 4.2] Section 4.2 defines local correlation as 'a separate correlation matrix for each 64-day quarter' but does not state whether this matrix is causal with respect to the prediction day. If the matrix for a 64-day block is computed using the full block, then for a prediction at day t+1 inside that block the matrix contains information from days after t, leaking future information into training and test. This would invalidate the local and dual rows of Tables 3 and 6 and affect the scope comparison (global vs local vs dual). The manuscript should specify the exact computation timeline or recompute all local/dual results using only past observations.
[Tables 3 and 6] Tables 3 and 6 report a single RMSE/MAE per configuration, with no standard deviations, no number of random seeds, and no description of how model selection across epochs was repeated. Given that several configurations differ by less than 5% (e.g., Spearman global RMSE 0.475 vs Spearman dual 0.472) and that training is stochastic, the claimed ranking of correlation metrics and scopes is not statistically supported. Please report mean±std over multiple runs and, where applicable, perform paired significance tests.
[Section 4.3.3] The only non-DGT baseline in Table 3 is a GRU. The conclusion in Section 4.3.3 that 'the DGT architecture has an inherent advantage in handling stock time series data' cannot be drawn from this comparison alone, because the DGT differs from the baseline in multiple dimensions (attention, graph prior, residual structure, normalization). A plain Transformer (spatial attention disabled) is already reported as 'DGT False' in Table 3, but it is still a DGT without spatial module; compare instead against a standard Transformer encoder with the same input projections and a GCN-GRU or similar static-graph model to isolate the contribution of differential graph attention.

minor comments (4)

[Section 3.4] The projection for values is written as `[K_t1; K_t2] = X_t W_K^T` twice; the second occurrence should be for V_t.
[Section 5.3] The text states that Kendall's Tau global achieves 'MAE of 0.098' for Cluster 0, but Table 6 reports 0.142959 for Cluster 0; the 0.098 figure corresponds to the Cluster 1 MAE. Please correct the inconsistency.
[Table 3] The row 'DGT True None None' is ambiguous: if the correlation scope is None, what does 'Use Spatial' mean? The ablation should clarify whether spatial attention is enabled with an uninformative prior or completely disabled.
[Table 6] The p-values in Table 6 are reported without stating the statistical test. Please identify the test (e.g., paired t-test) and the unit of analysis.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the reported DGT gains are empirical test-set results, with correlation matrices used as input features rather than fitted targets.

full rationale

The paper's central claims are empirical: Table 3 reports RMSE/MAE on a held-out test set after time-based train/validation/test partitioning (Section 4.1), and the best correlation scope and metric are selected on validation performance (Section 3.5). This is standard model selection, not a fitted input renamed as a prediction. The correlation matrices are input features, not the predicted quantity, and the prediction target (next-day normalized price) is not used to construct the graphs. The differential-attention formulas in Sections 2.3 and 3.4 are mutually inconsistent, and the Section 4.2 local-correlation description does not explicitly state causality, but these are architecture-identification and data-leakage concerns, not circular reductions. The citations most relevant to the mechanism are external: Ye et al. (2025) for Differential Transformer and K.X. Li (2025) for a related write-up; neither is a self-citation by the present authors, and no load-bearing argument reduces to an unverified self-citation. No equation in the paper is defined in terms of the result it is used to support, so there is no circular step.

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

The method depends on several standard ML hyperparameters and domain assumptions rather than fitted physical constants. The most consequential assumptions are the equivalence of the two differential attention formulas and the causal validity of local correlation matrices.

free parameters (4)

lambda_init = 0.2 for first layer
Initial offset for the learnable differential attention weight lambda; set empirically in Section 3.4.
learning_rate = 0.01 or 0.1 (selected by grid search)
Learning rate chosen from {0.01, 0.1} based on validation performance; Section 3.5.
number_of_clusters_k = 2
Chosen via silhouette and elbow methods in Section 5.1; affects cluster-level error analysis.
window_length = 64 trading days
Quarter-sized sliding windows for local correlation and model input; Section 4.1.

assumptions (5)

domain assumption Past z-score normalized closing prices contain predictive information for the next day's normalized price.
The entire forecasting task assumes short-term predictability in normalized prices; stated in Sections 1 and 4.
domain assumption Correlation matrices computed from price series are valid attention priors for cross-stock information propagation.
The DGT injects correlation matrices into attention as edge weights; Section 3.4.
domain assumption The Differential Transformer formulation of Ye et al. (2025) transfers to graph-structured financial inputs.
The paper adopts the two-stream differential attention and lambda formula from Ye et al. without re-deriving its validity for stock graphs; Section 3.4.
ad hoc to paper The two descriptions of differential graph attention in Sections 2.3 and 3.4 are equivalent.
Section 2.3 defines DiffAttention via Delta A_t added inside softmax, while Section 3.4 uses Hadamard multiplication with A_t and a two-stream softmax difference; the paper does not prove equivalence.
domain assumption K-means with k=2 yields meaningful stock groups for error analysis.
Silhouette and elbow support k=2, but the clusters are used downstream to draw conclusions; Section 5.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A Study of Dynamic Stock Relationship Modeling and S&P500 Price Forecasting Based on Differential Graph Transformer." pith.science (2026). https://pith.science/paper/JHVDSOYJ

@misc{pith2026250618717,
  author       = {Pith},
  title        = {Pith review of: A Study of Dynamic Stock Relationship Modeling and S&P500 Price Forecasting Based on Differential Graph Transformer},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JHVDSOYJ}},
  note         = {Machine review of arXiv:2506.18717}
}

read the original abstract

Stock price prediction is vital for investment decisions and risk management, yet remains challenging due to markets' nonlinear dynamics and time-varying inter-stock correlations. Traditional static-correlation models fail to capture evolving stock relationships. To address this, we propose a Differential Graph Transformer (DGT) framework for dynamic relationship modeling and price prediction. Our DGT integrates sequential graph structure changes into multi-head self-attention via a differential graph mechanism, adaptively preserving high-value connections while suppressing noise. Causal temporal attention captures global/local dependencies in price sequences. We further evaluate correlation metrics (Pearson, Mutual Information, Spearman, Kendall's Tau) across global/local/dual scopes as spatial-attention priors. Using 10 years of S&P 500 closing prices (z-score normalized; 64-day sliding windows), DGT with spatial priors outperformed GRU baselines (RMSE: 0.24 vs. 0.87). Kendall's Tau global matrices yielded optimal results (MAE: 0.11). K-means clustering revealed "high-volatility growth" and "defensive blue-chip" stocks, with the latter showing lower errors (RMSE: 0.13) due to stable correlations. Kendall's Tau and Mutual Information excelled in volatile sectors. This study innovatively combines differential graph structures with Transformers, validating dynamic relationship modeling and identifying optimal correlation metrics/scopes. Clustering analysis supports tailored quantitative strategies. Our framework advances financial time-series prediction through dynamic modeling and cross-asset interaction analysis.

Review history (2 revisions) →

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

STN-TGAT: Top-K Portfolio Construction via Prior-Guided Graph Attention with Learnable Soft-Threshold Sparsification
cs.LG 2026-07 conditional novelty 4.0 of 10

STN-TGAT, a Transformer plus graph attention model with an NMI relationship prior and learnable soft-threshold sparsification, shows higher backtested risk-adjusted returns than GRU, LSTM, and graph baselines on a Top...

Reference graph

Works this paper leans on

32 extracted references · 16 canonical work pages · cited by 1 Pith paper

[1]

China Funds Slash ETF Fees, Escalating Price War in Booming Market

Introduction 1.1 Research background and significance As more individuals and institutions around the world turn to the stock market for asset appreciation, it becomes increasingly apparent that actively managed equity funds struggle to outperform their benchmarks. Taking the U.S. market as an example, in a large -cap-dominated investment environment, mos...

work page 2018
[2]

black-box

Related works 2.1 Traditional Stock Prediction Methods In early research on stock price prediction, scholars mainly relied on statistical and classical machine learning models. These methods can generally be divided into two categories: time -series- based models and supervised learning based on feature engineering. First, time-series models such as ARIMA...

work page 2021
[3]

This approach overcomes the over -smoothing and scattered attention weights problems found in traditional Graph Convolutional Networks (GCNs)

Methodology 3.1 The basic structure of DGT The core of DGT lies in its ability to integrate spatial correlations and temporal dependencies, dynamically adjusting the attention mechanism to filter out market noise and uncover key stock relationships. This approach overcomes the over -smoothing and scattered attention weights problems found in traditional G...

work page 2019
[4]

Due to the delisting of Catalent (CTLT) and Marathon Oil (MRO) at the end of 2024, their historical data is incomplete

Research process and results 4.1 Dataset Selection and Preprocessing In this study, we selected the daily closing prices of the S&P 500 component stocks as our dataset, covering a 10-year period from March 2, 2015, to February 28, 2025. Due to the delisting of Catalent (CTLT) and Marathon Oil (MRO) at the end of 2024, their historical data is incomplete. ...

work page 2023
[5]

Discussion 5.1 Clustering Method Selection In stock market analysis, clustering stocks into different categories (e.g., technology, consumer) can significantly improve model interpretability and forecasting performance. The previous experiments discussed different types of stocks together, which li mits the ability to explain the prediction differences am...

work page
[6]

It offers a novel perspective for time-series forecasting and relationship modeling in financial markets

Conclusion 6.1 Main Research Conclusions This study proposes a dynamic stock relationship modeling framework based on the Differential Graph Transformer (DGT), significantly improving the prediction accuracy for the prices of S&P 500 constituent stocks. It offers a novel perspective for time-series forecasting and relationship modeling in financial market...

work page 2015
[7]

D., Edwards, T., Chapman, F., & Didio, N

Ganti, A., Gioia, D. D., Edwards, T., Chapman, F., & Didio, N. (2024). SPIVA U.S. Scorecard Year- End 2024

work page 2024
[8]

Chen, Y., Wei, Z., & Huang, X. (2018). Incorporating Corporation Relationship via Graph Convolutional Neural Networks for Stock Price Prediction. Proceedings of the 27th ACM International Conference on Information and Knowledge Management, 1655–1658. https://doi.org/10.1145/3269206.3269269

arXiv 2018

Show all 32 references

[9]

Cheng, D., Yang, F., Xiang, S., & Liu, J. (2022). Financial time series forecasting with multi- modality graph neural network. Pattern Recognition, 121, 108218. https://doi.org/10.1016/j.patcog.2021.108218

2022
[10]

China funds slash ETF fees, escalating price war in booming market. (2024). Reuters. https://www.reuters.com/markets/asia/china-funds-cut-etf-fees-escalating-price-war-booming-market- 2024-11-20/

2024
[11]

Determine the optimal value of K in K-Means Clustering – ML | GeeksforGeeks. (2025). https://www.geeksforgeeks.org/ml-determine-the-optimal-value-of-k-in-k-means-clustering/

2025
[12]

Ding, Q., Wu, S., Sun, H., Guo, J., & Guo, J. (2020). Hierarchical Multi-Scale Gaussian Transformer for Stock Movement Prediction. Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence, 4640–4646. https://doi.org/10.24963/ijcai.2020/640

2020 doi
[13]

Feng, F., Chen, H., He, X., Ding, J., Sun, M., & Chua, T.-S. (2019). Enhancing Stock Movement Prediction with Adversarial Training. Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence, 5843–5849. https://doi.org/10.24963/ijcai.2019/810

2019 doi
[14]

Qian, H., Zhou, H., Zhao, Q., Chen, H., Yao, H., Wang, J., Liu, Z., Yu, F., Zhang, Z., & Zhou, J. (2024). MDGNN: Multi-Relational Dynamic Graph Neural Network for Comprehensive and Dynamic Stock Investment Prediction. Proceedings of the AAAI Conference on Artificial Intelligen...

2024 doi
[15]

Jin, G., Liang, Y., Fang, Y., Shao, Z., Huang, J., Zhang, J., & Zheng, Y. (2023). Spatio-Temporal Graph Neural Networks for Predictive Learning in Urban Computing: A Survey (No. arXiv:2303.14483). arXiv. https://doi.org/10.48550/arXiv.2303.14483

2023 doi arXiv
[16]

Li, C., & Qian, G. (2023). Stock Price Prediction Using a Frequency Decomposition Based GRU Transformer Neural Network. Applied Sciences, 13(1), Article 1. https://doi.org/10.3390/app13010222

2023 doi
[17]

Li, K. X. (2025). Stock Market Forecasting with Differential Graph Transformer. Stanford CS224W: Machine Learning with Graphs. https://medium.com/stanford-cs224w/stock-market-forecasting-with- differential-graph-transformer-62d095ebc821

2025
[18]

Lu, W., Li, J., Wang, J., & Qin, L. (2021). A CNN-BiLSTM-AM method for stock price prediction. Neural Computing and Applications, 33(10), 4741–4753. https://doi.org/10.1007/s00521-020-05532- z

2021 doi
[19]

Ma, D., Yuan, D., Huang, M., & Dong, L. (2024). VGC-GAN: A multi-graph convolution adversarial network for stock price prediction. Expert Systems with Applications, 236, 121204. https://doi.org/10.1016/j.eswa.2023.121204

2024
[20]

A., & Krause, A

Moghimi, F., Johnson, R. A., & Krause, A. (2023). Rethinking Real Estate Pricing with Transformer Graph Neural Networks (T-GNN). 2023 International Conference on Machine Learning and Applications (ICMLA), 1405–1411. https://doi.org/10.1109/ICMLA58977.2023.00212

2023
[21]

Ye, T., Dong, L., Xia, Y., Sun, Y., Zhu, Y., Huang, G., & Wei, F. (2025). Differential Transformer (No. arXiv:2410.05258). arXiv. https://doi.org/10.48550/arXiv.2410.05258

2025 doi arXiv
[22]

Safak, A. N. (2024). The Elbow Method for the Optimal Value of K in K-means Clustering. Medium. https://medium.com/@ayse_nur_safak/elbow-method-for-optimal-value-of-k-in-k-means-clustering- e33a34e8f2a8

2024
[23]

Sawhney, R., Agarwal, S., Wadhwa, A., Derr, T., & Shah, R. R. (2021). Stock Selection via Spatiotemporal Hypergraph Attention Network: A Learning to Rank Approach. Proceedings of the AAAI Conference on Artificial Intelligence, 35(1), 497–504. https://doi.org/10.1609/aaai.v35i1.16127

2021 doi
[24]

Tian, H., Zhang, X., Zheng, X., & Zeng, D. D. (2023). Learning Dynamic Dependencies With Graph Evolution Recurrent Unit for Stock Predictions. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 53(11), 6705–6717. https://doi.org/10.1109/TSMC.2023.3284840

2023
[25]

N., Kaiser, Ł., & Polosukhin, I

Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., & Polosukhin, I. (2017). Attention is all you need. Proceedings of the 31st International Conference on Neural Information Processing Systems, 6000–6010

2017
[26]

Wang, C., Chen, Y., Zhang, S., & Zhang, Q. (2022). Stock market index prediction using deep Transformer model. Expert Systems with Applications, 208, 118128. https://doi.org/10.1016/j.eswa.2022.118128

2022
[27]

Xiang, S., Cheng, D., Shang, C., Zhang, Y., & Liang, Y. (2022). Temporal and Heterogeneous Graph Neural Network for Financial Time Series Prediction. Proceedings of the 31st ACM International Conference on Information & Knowledge Management, 3584–3593. https://doi.org/10.1145/...

2022
[29]

Yin, X., Yan, D., Almudaifer, A., Yan, S., & Zhou, Y. (2021). Forecasting Stock Prices Using Stock Correlation Graph: A Graph Convolutional Network Approach. 2021 International Joint Conference on Neural Networks (IJCNN), 1–8. https://doi.org/10.1109/IJCNN52387.2021.9533510

2021
[30]

Zhang, L., Aggarwal, C., & Qi, G.-J. (2017). 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. https://doi.org/10.1145/3097983.3098117

2017
[31]

Zheng, C., Fan, X., Wang, C., & Qi, J. (2019). GMAN: A Graph Multi-Attention Network for Traffic Prediction (No. arXiv:1911.08415). arXiv. https://doi.org/10.48550/arXiv.1911.08415

2019 doi arXiv
[32]

Zheng, Z., Shao, J., Zhu, J., & Shen, H. T. (2023). Relational Temporal Graph Convolutional Networks for Ranking-Based Stock Prediction. 2023 IEEE 39th International Conference on Data Engineering (ICDE), 123–136. https://doi.org/10.1109/ICDE55515.2023.00017

2023
[2025]

Spearman ranks second. In contrast, MI performs less effectively: for example, FANG shows an opposite trend to AAPL during the second half of December, suggesting that MI may capture more complex nonlinear dependencies that do not necessarily align with trend similarity. Unlik...

Pith tools

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

[1] [1]

China Funds Slash ETF Fees, Escalating Price War in Booming Market

Introduction 1.1 Research background and significance As more individuals and institutions around the world turn to the stock market for asset appreciation, it becomes increasingly apparent that actively managed equity funds struggle to outperform their benchmarks. Taking the U.S. market as an example, in a large -cap-dominated investment environment, mos...

work page 2018

[2] [2]

black-box

Related works 2.1 Traditional Stock Prediction Methods In early research on stock price prediction, scholars mainly relied on statistical and classical machine learning models. These methods can generally be divided into two categories: time -series- based models and supervised learning based on feature engineering. First, time-series models such as ARIMA...

work page 2021

[3] [3]

This approach overcomes the over -smoothing and scattered attention weights problems found in traditional Graph Convolutional Networks (GCNs)

Methodology 3.1 The basic structure of DGT The core of DGT lies in its ability to integrate spatial correlations and temporal dependencies, dynamically adjusting the attention mechanism to filter out market noise and uncover key stock relationships. This approach overcomes the over -smoothing and scattered attention weights problems found in traditional G...

work page 2019

[4] [4]

Due to the delisting of Catalent (CTLT) and Marathon Oil (MRO) at the end of 2024, their historical data is incomplete

Research process and results 4.1 Dataset Selection and Preprocessing In this study, we selected the daily closing prices of the S&P 500 component stocks as our dataset, covering a 10-year period from March 2, 2015, to February 28, 2025. Due to the delisting of Catalent (CTLT) and Marathon Oil (MRO) at the end of 2024, their historical data is incomplete. ...

work page 2023

[5] [5]

Discussion 5.1 Clustering Method Selection In stock market analysis, clustering stocks into different categories (e.g., technology, consumer) can significantly improve model interpretability and forecasting performance. The previous experiments discussed different types of stocks together, which li mits the ability to explain the prediction differences am...

work page

[6] [6]

It offers a novel perspective for time-series forecasting and relationship modeling in financial markets

Conclusion 6.1 Main Research Conclusions This study proposes a dynamic stock relationship modeling framework based on the Differential Graph Transformer (DGT), significantly improving the prediction accuracy for the prices of S&P 500 constituent stocks. It offers a novel perspective for time-series forecasting and relationship modeling in financial market...

work page 2015

[7] [7]

D., Edwards, T., Chapman, F., & Didio, N

Ganti, A., Gioia, D. D., Edwards, T., Chapman, F., & Didio, N. (2024). SPIVA U.S. Scorecard Year- End 2024

work page 2024

[8] [8]

Chen, Y., Wei, Z., & Huang, X. (2018). Incorporating Corporation Relationship via Graph Convolutional Neural Networks for Stock Price Prediction. Proceedings of the 27th ACM International Conference on Information and Knowledge Management, 1655–1658. https://doi.org/10.1145/3269206.3269269

arXiv 2018