{"paper_id":"2507.09739","title":"Enhancing Trading Performance Through Sentiment Analysis with Large Language Models: Evidence from the S&P 500","authors":["Haojie Liu","Zihan Lin","Randall R. Rojas"],"paper_date":"2025-07-13","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2507.09739v1","page_url":"https://thequant.space/flowcharts/enhancing-trading-performance-through-sentiment-analysis-wit/","scores":{"empirical_rigor":6,"math_complexity":4.5,"quadrant":"Street Traders"},"proposes":{"status":"available","statement":"The paper proposes a hybrid trading strategy that integrates real-time sentiment analysis from financial news using GPT-2 and FinBERT with technical indicators and time‑series models (ARIMA, ETS) to improve S&P 500 trading performance.","quote":"This study integrates real-time sentiment analysis from financial news, GPT-2 and FinBERT, with technical indicators and time-series models like ARIMA and ETS to optimize S&P 500 trading strategies.","summary":"This study integrates real-time sentiment analysis from financial news, GPT-2 and FinBERT, with technical indicators and time-series models like ARIMA and ETS to optimize S&P 500 trading strategies. By merging sentiment data with momentum and trend-based metrics, including a benchmark buy-and-hold…","strategy":"Sentiment-driven trading strategy combining GPT-2 (or FinBERT) with VW MACD"},"assumptions":{"status":"available","items":[{"assumption":"The simulation assumes zero transaction costs and slippage.","type":"execution","stated":"explicit","quote":"For simplicity, this simulation assumes zero transaction costs and slippage."},{"assumption":"Sentiment models treat each headline equally, ignoring article length, author credibility, or source‑specific biases.","type":"model","stated":"explicit","quote":"Our sentiment models treat each headline equally and do not account for article length, author credibility, or source-specific biases."}]},"datasets":{"status":"available","items":[{"name":"Financial news articles","provider":"Wall Street Journal, Barron, Benzinga, MarketWatch, Dow Jone","kind":"text-news","access":"public","period":null,"universe":null,"quote":"In this study, we utilize two key datasets: financial news articles collected from sources such as the Wall Street Journal (WSJ), Barron, Benzinga, MarketWatch, and Dow Jones, along with daily stock prices for the S&P 500 index."},{"name":"S&P 500 daily price data","provider":"Yahoo Finance","kind":"market-prices","access":"public","period":"August 08, 2019 to August 07, 2024","universe":"S&P 500 index","quote":"We downloaded this data from Yahoo Finance, capturing daily prices from August 08, 2019 to August 07, 2024.","known_id":"yahoo-finance"}],"known":[{"id":"yahoo-finance","name":"Yahoo Finance","provider":"Yahoo","kind":"market-prices","access":"public","quote":"We downloaded this data from Yahoo Finance, capturing daily prices from August 08, 2019 to August 07, 2024.","mentions":1,"used_in":1}],"sample":{"start":null,"end":null,"market":"S&P 500","asset_class":"equities","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"no","headline_net_of_costs":false,"detail":null,"quote":"For simplicity, this simulation assumes zero transaction costs and slippage.","source":"front_matter","rule":"R3","candidates":{"front_matter":"no","evidence_card":"no","evidence_card_quoted":true},"card_detail":null,"card_quote":"For simplicity, this simulation assumes zero transaction costs and slippage.","meaning":"a trading or execution result is reported without costs"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"The hybrid strategy outperformed the baseline buy‑and‑hold, delivering a top‑performing return of 5.77%.","kind":"performance","quote":"The top-performing version delivered a 5.77% return, demonstrating the value that can be had."},{"finding":"GPT‑2 combined with VW MACD and ARIMA achieved the highest overall return of 4.39% for Dow Jones data.","kind":"performance","quote":"GPT-2 achieves the highest return overall, 4.39%, when paired with VW MACD ARIMA for Dow Jones, surpassing FinBERT’s 0.52%."},{"finding":"Combining sentiment‑driven insights with traditional models improves trading performance.","kind":"performance","quote":"Results show that combining sentiment-driven insights with traditional models improves trading performance."}],"reported_performance":{"strategy":"Sentiment-driven trading strategy combining GPT-2 (or FinBERT) with VW MACD","sharpe":null,"sharpe_benchmark":null,"annual_return_pct":null,"total_return_pct":5.77,"max_drawdown_pct":null},"out_of_sample":{"method":"holdout","headline_out_of_sample":true}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":17,"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":"2026-11-14T17:32:51Z"},"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/"}