{"paper_id":"2502.01574","title":"An End-To-End LLM Enhanced Trading System","authors":["Ziyao Zhou","Ronitt Mehra"],"paper_date":"2025-02-03","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2502.01574v1","page_url":"https://thequant.space/flowcharts/an-end-to-end-llm-enhanced-trading-system/","scores":{"empirical_rigor":6,"math_complexity":3.5,"quadrant":"Street Traders"},"proposes":{"status":"available","statement":"The paper proposes an end-to-end trading system that uses FinGPT for real-time market sentiment analysis combined with technical indicators to generate trading signals.","quote":"This project introduces an end-to-end trading system that leverages Large Language Models (LLMs) for real-time market sentiment analysis.","summary":"This project introduces an end-to-end trading system that leverages Large Language Models (LLMs) for real-time market sentiment analysis. By synthesizing data from financial news and social media, the system integrates sentiment-driven insights with technical indicators to generate actionable…","strategy":"sentiment-enhanced trading strategy"},"assumptions":{"status":"available","items":[{"assumption":"Trading decisions are based on sentiment signal strength, allocating 10% or 15% of initial cash per trade.","type":"execution","stated":"explicit","quote":"trading decisions are influenced by sentiment signals, where the signal strength determines the trade size (10%or15% of initial cash)"},{"assumption":"Positions are closed at the last available price and profit‑taking is triggered by position flipping.","type":"execution","stated":"explicit","quote":"position flipping (from long to short or vice versa) triggers profit-taking, and positions are closed at the last available price"}]},"datasets":{"status":"available","items":[{"name":"Kaggle Financial Sentiment Analysis dataset","provider":"Kaggle","kind":"text-news","access":"public","period":null,"universe":null,"quote":"The dataset chosen for this benchmarking exercise, sourced from Kaggle’s Financial Sentiment Analysis dataset, contains financial news headlines and their corresponding sentiment labels.","known_id":"kaggle"}],"known":[{"id":"reddit","name":"Reddit","provider":"Reddit","kind":"text-news","access":"public","quote":"Prices and volumes are continu-ously ingested and logged into a buffer for further analysis. • Financial Text Data: – News Articles: Relevant financial articles men-tioning the selected tickers are fetched via the News API. – Reddit Posts: Posts and comments from the WallStreetBets subreddit are retrieved using the PRAW API.","mentions":4,"used_in":1},{"id":"kaggle","name":"Kaggle","provider":"Kaggle","kind":"other","access":"public","quote":"TABLE I PERFORMANCE COMPARISON OF MODELS Model Accuracy Precision Recall F1-Score FinGPT 0.7462 0.7675 0.7642 0.7488 IBM Granite 3.0 0.5861 0.6942 0.5861 0.6207 Meta LLAMA 3.1 0.6565 0.6657 0.6565 0.6440 The dataset chosen for this benchmarking exercise, sourced from Kaggle’s Financial Sentiment Analysis dataset, contains financial news headlines and their corresponding sentiment labels.","mentions":1,"used_in":1}],"sample":{"start":null,"end":null,"market":null,"asset_class":"equities","frequency":"intraday"},"data_resolution":"intraday-bars"},"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":"FinGPT outperforms IBM Granite 3.0 and Meta LLaMA 3.1 on sentiment classification with accuracy 0.7462, precision 0.7675, recall 0.7642, and F1 0.7488.","kind":"predictive-accuracy","quote":"FinGPT 0.7462 0.7675 0.7642 0.7488"},{"finding":"Sentiment‑integrated SMA Crossover strategy for TSLA improves Sharpe ratio from 0.34 to 3.47.","kind":"performance","quote":"For example, the SMA Crossover strategy for TSLA improves from 0.34 to 3.47"},{"finding":"Sentiment‑integrated SMA Crossover strategy for TSLA raises win ratio from 32.2% to 57.0%.","kind":"performance","quote":"For TSLA, the SMA Crossover win ratio jumps from 32.2% to 57.0%"}],"reported_performance":{"strategy":"sentiment-enhanced trading strategy","sharpe":3.47,"sharpe_benchmark":null,"annual_return_pct":null,"total_return_pct":null,"max_drawdown_pct":null},"out_of_sample":{"method":"unclear","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":26,"citations":{"cited_by_in_corpus":1,"by_independent_authors":1,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":1,"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":"2026-11-14T17:45:56Z"},"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/"}