{"paper_id":"2412.19245","title":"Sentiment trading with large language models","authors":["Kemal Kirtac","Guido Germano"],"paper_date":"2024-12-26","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2412.19245v1","page_url":"https://thequant.space/flowcharts/sentiment-trading-with-large-language-models/","scores":{"empirical_rigor":6,"math_complexity":6.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes using large language models for sentiment analysis of U.S. financial news and building long‑short trading strategies based on the derived sentiment scores.","quote":"We analyse the performance of the large language models (LLMs) OPT, BERT, and FinBERT, alongside the traditional Loughran-McDonald dictionary, in the sentiment analysis of 965,375 U.S. financial news articles from 2010 to 2023.","summary":"We investigate the efficacy of large language models (LLMs) in sentiment analysis of U.S. financial news and their potential in predicting stock market returns. We analyze a dataset comprising 965,375 news articles that span from January 1, 2010, to June 30, 2023; we focus on the performance of…","strategy":"long-short OPT strategy"},"assumptions":{"status":"available","items":[{"assumption":"Trading incurs a flat transaction cost of 10 basis points per trade.","type":"execution","stated":"explicit","quote":"To make our simulation more aligned with actual trading conditions, we included a transaction cost of 10 basis points for each trade, accounting for the typical costs traders would encounter in the market."},{"assumption":"Portfolios are rebalanced daily using market‑value weighting and are self‑financing long‑short strategies.","type":"execution","stated":"explicit","quote":"The long‑short portfolios are self‑financing strategies that simultaneously involve taking long positions in stocks with the highest 20% sentiment scores and short positions in stocks with the lowest 20% sentiment scores."},{"assumption":"News articles are filtered for uniqueness using a cosine similarity threshold of 0.8 within the prior five business days.","type":"data","stated":"explicit","quote":"We avoid redundancy by using a novelty score based on the similarity between articles: if a new article is too similar to an older article published within the past 20 days according to a cosine similarity score of 0.8 or more, we exclude it."},{"assumption":"Only news linked to individual stocks with available three‑day returns are included.","type":"data","stated":"explicit","quote":"We only consider news articles related to individual stocks with available three‑day returns."},{"assumption":"Trading decisions are timed to news releases: pre‑6 am news trades at market open, 6 am‑4 pm news trades same‑day close, post‑4 pm news trades next day open.","type":"execution","stated":"explicit","quote":"For news appearing between 6 am and 4 pm, we initiate a trade with closing prices of the same day and exit the trade the next trading day. Any news coming in after 4 pm was used for trades at the start of the next trading day."}]},"datasets":{"status":"available","items":[{"name":"CRSP daily stock file","provider":"CRSP","kind":"market-prices","access":"commercial","period":"2010-01-01 to 2023-06-30","universe":"U.S. stocks listed on AMEX, NASDAQ, NYSE","quote":"In our research, we primarily use two datasets: one from the Center for Research in Security Prices (CRSP) that includes daily stock returns, and another from Refinitiv with global news.","known_id":"crsp"},{"name":"Refinitiv news dataset","provider":"Refinitiv","kind":"text-news","access":"commercial","period":"2010-01-01 to 2023-06-30","universe":"U.S. companies covered in news articles","quote":"The news data from Refinitiv comprises detailed articles and quick alerts, focusing on companies based in the U.S.","known_id":"refinitiv"},{"name":"Filtered news articles","provider":"Refinitiv (filtered)","kind":"text-news","access":"commercial","period":"2010-01-01 to 2023-06-30","universe":"965,375 unique U.S. stock‑specific articles after filters","quote":"After applying our filters, we were left with 965,375 articles. Our sample dataset is summarized in Table 1.","known_id":"refinitiv"}],"known":[{"id":"refinitiv","name":"Refinitiv / LSEG (Datastream, Eikon, Tick History)","provider":"LSEG","kind":"market-prices","access":"commercial","quote":"The news data from Refinitiv comprises detailed articles and quick alerts, focusing on companies based in the U.S. The CRSP data provides daily return information for companies that trade on major U.S. stock exchanges.","mentions":4,"used_in":1},{"id":"crsp","name":"CRSP","provider":"Center for Research in Security Prices","kind":"market-prices","access":"commercial","quote":"The news data from Refinitiv comprises detailed articles and quick alerts, focusing on companies based in the U.S. The CRSP data provides daily return information for companies that trade on major U.S. stock exchanges.","mentions":3,"used_in":1},{"id":"financial-phrasebank","name":"Financial PhraseBank","provider":"Malo et al.","kind":"benchmark","access":"public","quote":"Huang et al. (2023) demonstrated this adaptability by fine-tuning a BERT model for classification tasks using the Financial PhraseBank dataset, initially compiled by Malo et al. (2014).","mentions":1,"used_in":1}],"sample":{"start":"2021-08","end":"2023-07","market":"U.S. stocks","asset_class":"equities","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"yes","headline_net_of_costs":null,"detail":"10 basis points per trade, applied to each side of the transaction.","quote":"To make our simulation more aligned with actual trading conditions, we included a transaction cost of 10 basis points for each trade, accounting for the typical costs traders would encounter in the market.","source":"evidence_card","rule":"R2","candidates":{"front_matter":"unclear","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"10 basis points per trade, applied to each side of the transaction.","card_quote":"To make our simulation more aligned with actual trading conditions, we included a transaction cost of 10 basis points for each trade, accounting for the typical costs traders would encounter in the market.","meaning":"costs are part of the reported headline results"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"OPT achieves the highest sentiment prediction accuracy of 74.4%, outperforming BERT (72.5%) and FinBERT (72.2%).","kind":"predictive-accuracy","quote":"Specifically, OPT, which is a GPT‑3 based LLM, shows the highest accuracy in sentiment prediction with an accuracy of 74.4%, slightly ahead of BERT (72.5%) and FINBERT (72.2%)."},{"finding":"The long‑short OPT strategy yields a Sharpe ratio of 3.05, far above BERT (2.11), FinBERT (2.07) and the Loughran‑McDonald dictionary (1.23).","kind":"performance","quote":"In portfolio performance, the long‑short OPT strategy excels with a Sharpe ratio of 3.05, compared to 2.11 for BERT and 2.07 for FINBERT long‑short strategies. Strategies based on the Loughran‑McDonald dictionary yield the lowest Sharpe ratio of 1.23."},{"finding":"From August 2021 to July 2023 the OPT long‑short strategy generated a 355% cumulative gain.","kind":"performance","quote":"From August 2021 to July 2023, this strategy produces an impressive 355% gain, outperforming other strategies and traditional market portfolios."},{"finding":"Regression coefficients for OPT sentiment scores are positive and significant (0.274 and 0.254) in predicting next‑day returns.","kind":"statistical","quote":"Regression analyses highlight a robust positive impact of OPT model scores on next‑day stock returns, with coefficients of 0.274 and 0.254 in different model specifications."},{"finding":"The Loughran‑McDonald dictionary model shows low predictive power (50.1% accuracy) and no significant relationship with stock returns.","kind":"negative","quote":"In contrast, the Loughran‑McDonald dictionary model demonstrates considerably lower effectiveness with only 50.1% accuracy"}],"reported_performance":{"strategy":"long-short OPT strategy","sharpe":3.05,"sharpe_benchmark":1.23,"annual_return_pct":null,"total_return_pct":355,"max_drawdown_pct":18.57},"out_of_sample":{"method":"holdout","headline_out_of_sample":true}},"independent_evidence":{"status":"available","supports":[],"contradicts":[],"qualifies":[],"compared_with":45,"citations":{"cited_by_in_corpus":12,"by_independent_authors":11,"how_cited":{"used_as_baseline":0,"uses_method_or_data":1,"extends":0,"discusses_findings":0,"mentions":10,"unclassified":1},"independent_papers":[{"paper":{"paper_id":"2511.13614","title":"Market-Dependent Communication in Multi-Agent Alpha Generation","paper_date":"2025-11-17","authors":["Jerick Shi","Burton Hollifield"],"empirical_rigor":6,"page_url":"https://thequant.space/flowcharts/market-dependent-communication-in-multi-agent-alpha-generati/"},"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":"2026-11-14T16:37:04Z"},"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/"}