{"paper_id":"2510.04357","title":"From News to Returns: A Granger-Causal Hypergraph Transformer on the Sphere","authors":["Anoushka Harit","Zhongtian Sun","Jongmin Yu"],"paper_date":"2025-10-05","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2510.04357v1","page_url":"https://thequant.space/flowcharts/from-news-to-returns-a-granger-causal-hypergraph-transforme/","scores":{"empirical_rigor":7,"math_complexity":8.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes the Causal Sphere Hypergraph Transformer (CSHT), a novel architecture that combines Granger‑causal hypergraph structure, Riemannian geometry, and causally masked Transformer attention for interpretable financial time‑series forecasting.","quote":"We propose the Causal Sphere Hypergraph Transformer (CSHT), a novel architecture for interpretable financial time-series forecasting that unifies Granger-causal hypergraph structure, Riemannian geometry, and causally masked Transformer attention.","summary":"We propose the Causal Sphere Hypergraph Transformer (CSHT), a novel architecture for interpretable financial time-series forecasting that unifies \\emph{“Granger-causal hypergraph structure”}, \\emph{“Riemannian geometry”}, and \\emph{“causally masked Transformer attention”}. CSHT models the…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Granger causality tests are performed with a maximum lag of K=5 to detect multi‑source influences on each return variable.","type":"statistical","stated":"explicit","quote":"We apply multivariate Granger causality tests with a maximum lag of"},{"assumption":"Only stocks with complete trading history are retained, resulting in 450 S&P 500 constituents.","type":"data","stated":"explicit","quote":"We focus on stocks in the S&P 500 index between 2018 and 2023. We retain 450 actively traded stocks after filtering out those with incomplete trading history."},{"assumption":"Features are z‑normalised using training‑set statistics.","type":"data","stated":"explicit","quote":"All features are z-normalised using training-set statistics."},{"assumption":"Attention masks are angular and preserve temporal directionality and geometric consistency.","type":"model","stated":"explicit","quote":"Attention is constrained via angular masks that preserve both temporal directionality and geometric consistency."},{"assumption":"Evaluation uses a non‑overlapping temporal split (training 2018‑2020, validation 2021, test 2022‑mid‑2023).","type":"execution","stated":"explicit","quote":"We use a non-overlapping temporal split: • Training: 2018–2020; • Validation: 2021; • Test: 2022–mid-2023."}]},"datasets":{"status":"available","items":[{"name":"Yahoo Finance historical price data","provider":"Yahoo Finance","kind":"market-prices","access":"public","period":"2018–2023","universe":"450 S&P 500 stocks","quote":"Historical prices and volumes are obtained from Yahoo Finance1.","known_id":"yahoo-finance"},{"name":"Twitter Financial News Sentiment dataset","provider":"HuggingFace (zeroshot/twitter-financial-news-sentiment)","kind":"text-news","access":"public","period":null,"universe":null,"quote":"We use the Twitter Financial News Sentiment dataset2, which includes over 500,000 financial tweets labeled as positive, negative, or neutral.","known_id":"twitter"}],"known":[{"id":"twitter","name":"Twitter / X","provider":"X Corp.","kind":"text-news","access":"public","quote":"We use the Twitter Financial News Sentiment dataset2, which includes over 500,000 financial tweets labeled as positive, negative, or neutral.","mentions":1,"used_in":1},{"id":"yahoo-finance","name":"Yahoo Finance","provider":"Yahoo","kind":"market-prices","access":"public","quote":"Historical prices and volumes are obtained from Yahoo Finance1.","mentions":1,"used_in":1}],"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":"CSHT achieves the strongest results across all metrics, reducing MAE to 0.0162 (13.4% lower than the best baseline), improving regime classification accuracy to 74.6% (3.3 points higher), and attaining the highest NDCG@10 of 0.683.","kind":"performance","quote":"CSHT achieves the strongest results across all metrics. It reduces MAE by 13.4% relative to the strongest baseline (FinGAT), improves classification accuracy by 3.3 percentage points, and attains the highest NDCG@10—demonstrating superior ranking under volatility."},{"finding":"CSHT shows substantially higher causal alignment, with 66.4% of attention mass aligned to Granger‑causal edges versus 21.3% for the next best interpretable baseline.","kind":"statistical","quote":"CSHT exhibits substantially higher alignment with validated causal structure."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":21,"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":"2026-11-14T17:25:38Z"},"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/"}