{"paper_id":"2502.21206","title":"Chronologically Consistent Large Language Models","authors":["Songrun He","Linying Lv","Asaf Manela","Jimmy Wu"],"paper_date":"2025-02-28","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2502.21206v3","page_url":"https://thequant.space/flowcharts/chronologically-consistent-large-language-models/","scores":{"empirical_rigor":6,"math_complexity":3,"quadrant":"Street Traders"},"proposes":{"status":"available","statement":"The paper introduces ChronoBERT and ChronoGPT, chronologically consistent large language models trained only on text available at each point in time.","quote":"we overcome this challenge by training a suite of chronologically consistent large language models, ChronoBERT and ChronoGPT, which incorporate only the text data that would have been available at each point in time.","summary":"Large language models are increasingly used in social sciences, but their training data can introduce lookahead bias and training leakage. A good chronologically consistent language model requires efficient use of training data to maintain accuracy despite time-restricted data. Here, we overcome…","strategy":"ChronoGPTRealtime"},"assumptions":{"status":"available","items":[{"assumption":"Market efficiency tests assume that prices reflect only known facts at the time.","type":"market","stated":"explicit","quote":"Market efficiency tests (Fama, 1970) assume that prices reflect only known facts at the time."}]},"datasets":{"status":"available","items":[{"name":"Dow Jones Newswire dataset","provider":"Dow Jones (subscribed by major institutional investors)","kind":"text-news","access":"commercial","period":null,"universe":"firm-specific news headlines within trading day window","quote":"... For each firm-day observation, we aggregate all news headlines related to the firm within the trading day window—spanning from 4:00 p.m. EST on day t −1 to 4:00 p.m. EST on day t—and treat the combined text as the firm’s textual information."}],"known":[],"sample":{"start":"2008-01","end":"2023-07","market":"US stocks","asset_class":"equities","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"no","headline_net_of_costs":false,"detail":null,"quote":null,"source":"front_matter","rule":"R1","candidates":{"front_matter":"no","evidence_card":"unclear","evidence_card_quoted":false},"meaning":"a trading or execution result is reported without costs"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"ChronoBERTRealtime and ChronoGPTRealtime generate Sharpe ratios of 4.80 and 4.92, outperforming GPT-2 XL, ModernBERT, BERT, FinBERT and StoriesLM.","kind":"performance","quote":"In this news return prediction setting, the H-L portfolios from ChronoBERTRealtime and ChronoGPTRealtime generate Sharpe ratios of 4.80 and 4.92, outperforming GPT-2 XL, ModernBERT, BERT, FinBERT and StoriesLM."},{"finding":"ChronoBERT and ChronoGPT have comparable Sharpe ratios to Llama 3.1 with no statistically significant differences (p‑value 0.315 and 0.547).","kind":"performance","quote":"Comparing the investment performance of ChronoBERT and ChronoGPT against the Llama 3.1 model, we find that ChronoBERT and ChronoGPT generate comparable Sharpe ratios, with no statistically significant differences (p-value of 0.315 and 0.547)."},{"finding":"Lookahead bias appears modest in the stock return forecasting application.","kind":"statistical","quote":"A somewhat surprising finding is that lookahead bias appears to be modest in this stock return forecasting application."},{"finding":"ChronoBERT and ChronoGPT match or surpass the language comprehension abilities of BERT and GPT-2 while achieving performance comparable to the much larger Llama 3.1 model.","kind":"predictive-accuracy","quote":"Our results show that ChronoBERT and ChronoGPT match or surpass the language comprehension abilities of BERT and GPT-2 while generating performance comparable to the much larger Llama 3.1 model in asset pricing applications."}],"reported_performance":{"strategy":"ChronoGPTRealtime","sharpe":4.92,"sharpe_benchmark":null,"annual_return_pct":null,"total_return_pct":null,"max_drawdown_pct":null},"out_of_sample":{"method":"walk-forward","headline_out_of_sample":true}},"independent_evidence":{"status":"available","supports":[],"contradicts":[],"qualifies":[],"compared_with":41,"citations":{"cited_by_in_corpus":13,"by_independent_authors":12,"how_cited":{"used_as_baseline":0,"uses_method_or_data":2,"extends":0,"discusses_findings":1,"mentions":7,"unclassified":3},"independent_papers":[{"paper":{"paper_id":"2511.18578","title":"Re(Visiting) Time Series Foundation Models in Finance","paper_date":"2025-11-23","authors":["Eghbal Rahimikia","Hao Ni","Weiguan Wang"],"empirical_rigor":8.5,"page_url":"https://thequant.space/flowcharts/revisiting-time-series-foundation-models-in-finance/"},"how":"discusses_findings","quote":"Their findings indicate that look-ahead bias can be reduced while maintaining strong performance in both language and financial prediction tasks."},{"paper":{"paper_id":"2512.23847","title":"A Test of Lookahead Bias in LLM Forecasts","paper_date":"2025-12-29","authors":["Zhenyu Gao","Wenxi Jiang","Yutong Yan"],"empirical_rigor":7,"page_url":"https://thequant.space/flowcharts/a-test-of-lookahead-bias-in-llm-forecasts/"},"how":"uses_method_or_data","quote":null},{"paper":{"paper_id":"2410.23447","title":"Continuous Risk Factor Models: Analyzing Asset Correlations through Energy Distance","paper_date":"2024-10-30","authors":["Marcus Gawronsky","Chun-Sung Huang"],"empirical_rigor":2.5,"page_url":"https://thequant.space/flowcharts/continuous-risk-factor-models-analyzing-asset-correlations/"},"how":"uses_method_or_data","quote":null}],"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":"2026-11-14T17:43:59Z"},"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/"}