{"paper_id":"2412.04034","title":"Dynamic Graph Representation with Contrastive Learning for Financial Market Prediction: Integrating Temporal Evolution and Static Relations","authors":["Yunhua Pei","Jin Zheng","John Cartlidge"],"paper_date":"2024-12-05","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2412.04034v1","page_url":"https://thequant.space/flowcharts/dynamic-graph-representation-with-contrastive-learning-for-f/","scores":{"empirical_rigor":8,"math_complexity":7,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes the Dynamic Graph Representation with Contrastive Learning (DGRCL) framework, which integrates dynamic temporal graph learning and static relational graph learning via an Embedding Enhancement (EE) module and a Contrastive Constrained Training (CCT) module.","quote":"To address this issue, we propose the Dynamic Graph Representation with Contrastive Learning (DGRCL) framework, which integrates dynamic and static graph relations to improve the accuracy of stock trend prediction. Our framework introduces two key components: the Embedding Enhancement (EE) module and the Contrastive Co","summary":"Temporal Graph Learning (TGL) is crucial for capturing the evolving nature of stock markets. Traditional methods often ignore the interplay between dynamic temporal changes and static relational structures between stocks. To address this issue, we propose the Dynamic Graph Representation with…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Dynamic edges are constructed adaptively following Zipf’s law.","type":"model","stated":"explicit","quote":"In embedding enhancement, we adaptively construct dynamic edges that follow Zipf’s law."},{"assumption":"The trade‑off parameter λ balances prediction loss and contrastive loss.","type":"model","stated":"explicit","quote":"The final form of the loss function is shown below, where λ is a trade‑off parameter."},{"assumption":"Sector‑industry relationships adequately capture static relational constraints among stocks.","type":"data","stated":"explicit","quote":"Recognizing that stocks within the same industry tend to be similarly affected by industry‑wide prospects, we gather data on the sector‑industry relationships among stocks."},{"assumption":"Training, validation and test splits are 0.65, 0.1, and 0.25 of the total trading days respectively.","type":"data","stated":"explicit","quote":"We follow the setup of (Pareja et al., 2020) and divide the two datasets into training, validation, and testing sets in the proportions of 0.65, 0.1, and 0.25, respectively."},{"assumption":"OHLCV daily values are sufficient features for stock prediction.","type":"data","stated":"explicit","quote":"OHLCV are daily values for each stock."}]},"datasets":{"status":"none_found","items":[],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"daily"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"forecast-accuracy","findings":[{"finding":"DGRCL achieves the highest classification accuracy on the NASDAQ dataset (53.06%) and NYSE dataset (54.07%), outperforming all baselines.","kind":"performance","quote":"On the NYSE dataset, DGRCL similarly outperforms other methods, with an accuracy of 54.07%."},{"finding":"DGRCL attains the top F1 scores of 66.53% (NASDAQ) and 67.53% (NYSE), and strong MCC values of 9.32×10⁻³ and 27.51×10⁻³ respectively.","kind":"predictive-accuracy","quote":"and a substantial MCC of 27.51, marking the highest recorded metrics across all models."},{"finding":"Training time is dominated by the CCT module, which accounts for around 90% of total runtime (≈414 s on NASDAQ, ≈506 s on NYSE).","kind":"other","quote":"The full DGRCL model, which includes these enhancements, requires 414.05s on NASDAQ and 505.54s on NYSE. Notably, adding the CCT unit significantly increases the training time, accounting for around 90% of the total run-time."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":32,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":1,"extracted_at":"2026-11-14T17:50:52Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}