{"paper_id":"2311.15180","title":"Benchmarking Large Language Model Volatility","authors":["Boyang Yu"],"paper_date":"2023-11-26","paper_type":"dataset-benchmark","paper_url":"https://arxiv.org/pdf/2311.15180v1","page_url":"https://thequant.space/flowcharts/benchmarking-large-language-model-volatility/","scores":{"empirical_rigor":6,"math_complexity":2.5,"quadrant":"Street Traders"},"proposes":{"status":"available","statement":"The paper proposes to analyze language processes through their generated samples to quantify LLM uncertainty without probability scores.","quote":"We propose to analyse language processes through their generated samples, because black-box LLMs do not explicitly return probability scores[7, 8].","summary":"The impact of non-deterministic outputs from Large Language Models (LLMs) is not well examined for financial text understanding tasks. Through a compelling case study on investing in the US equity market via news sentiment analysis, we uncover substantial variability in sentence-level sentiment…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The analysis is performed in a zero‑shot setup with no additional training.","type":"model","stated":"explicit","quote":"We study the models in a zero-shot setup, i.e., without providing additional training where we present text to models and ask them to provide a response containing sentiment labels."},{"assumption":"A daily long‑short rebalancing strategy is used for portfolio construction.","type":"execution","stated":"explicit","quote":"We adopt a long-short strategy to adjust holding positions on a daily basis."},{"assumption":"Zero temperature is favored for reproducibility but does not guarantee determinism.","type":"model","stated":"explicit","quote":"Published studies favors zero temperature to maximize reproducibility[7, 9, 14, 15]. Note that a temperature of 0 has its downsides. First, it does not guarantee determinism since small numerical discrepancies in floating precision and distributed computation can cause deviations in the output."}]},"datasets":{"status":"available","items":[{"name":"EODHD (Unicorn Data Services)","provider":"Unicorn Data Services","kind":"market-prices","access":"commercial","period":"October 2021 to September 2023","universe":"S&P 500 companies","quote":"We utilize two primary data sources for our analysis: Unicorn Data Services (EOD Historica lData | EODHD) for historical price data and news headlines..."},{"name":"Aylien news API","provider":"Aylien","kind":"text-news","access":"commercial","period":"October 2021 to September 2023","universe":"S&P 500 companies","quote":"serves as an additional data source for collecting news headlines from other sources, namely MSN, CNN, Seeking Alpha and Fox Business."}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"no","headline_net_of_costs":null,"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":"LLM sentiment scores are volatile across temperature settings, leading to significant variation in portfolio returns.","kind":"performance","quote":"Fig. 2. A long-short strategy show volatile returns using LLM sentiment scores. The accumulated PnLs are plotted for two settings, low temperature (t=0.0) and high temperature (t=1.0)."},{"finding":"Higher temperature settings increase return volatility and affect Sharpe ratios; e.g., LLaMA at t=0.0 yields 8.51 % ± 0.25 return and Sharpe 1.41 ± 0.04, while at t=1.0 returns 8.82 % ± 2.72 and Sharpe 1.46 ± 0.45.","kind":"performance","quote":"Table 2. Strategy performance is affected by volatile sentiment scores."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":9,"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":"2026-11-14T18:18:43Z"},"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/"}