{"paper_id":"2412.07587","title":"A Hype-Adjusted Probability Measure for NLP Stock Return Forecasting","authors":["Zheng Cao","Helyette Geman"],"paper_date":"2024-12-10","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2412.07587v6","page_url":"https://thequant.space/flowcharts/a-hype-adjusted-probability-measure-for-nlp-stock-return-for/","scores":{"empirical_rigor":6,"math_complexity":6.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper introduces a Hype-Adjusted Probability Measure for NLP-based stock return and volatility forecasting, with a novel sentiment score equation to capture intraday news impact.","quote":"This article introduces a Hype-Adjusted Probability Measure in the context of a new Natural Language Processing (NLP) approach for stock return and volatility forecasting.","summary":"This article introduces a Hype-Adjusted Probability Measure in the context of a new Natural Language Processing (NLP) approach for stock return and volatility forecasting. A novel sentiment score equation is proposed to represent the impact of intraday news on forecasting next-period stock return…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The model assumes three states of nature (Up, Level, Down) for the underlying probability space.","type":"model","stated":"explicit","quote":"We suppose that there are 3 states of nature ω1, ω2, ω3 in the universe Ω, labeled as Up, Level, and Down."}]},"datasets":{"status":"available","items":[{"name":"LSEG and Eikon API news dataset","provider":"LSEG/Eikon","kind":"text-news","access":"commercial","period":null,"universe":"30 semiconductor stocks","quote":"Primary data sources for this study were obtained from LSEG and the Eikon API [11], which provide access to over half a million news articles related to 30 semiconductor stocks.","known_id":"refinitiv"}],"known":[{"id":"refinitiv","name":"Refinitiv / LSEG (Datastream, Eikon, Tick History)","provider":"LSEG","kind":"market-prices","access":"commercial","quote":"Primary data sources for this study were obtained from LSEG and the Eikon API [11], which provide access to over half a million news articles related to 30 semiconductor stocks.","mentions":4,"used_in":2}],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"daily"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R1","candidates":{"front_matter":"n/a","evidence_card":"unclear","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"forecast-accuracy","findings":[{"finding":"Validation accuracy for market return direction improves from 51.7% (baseline) to 70.0% (LDA model) and further to 78.3% with optimized scores.","kind":"performance","quote":"In the prediction of market return direction, validation accuracy improves from 51.7% in the baseline model to 70.0% with a LDA model (as presented by Deveikyte et al [7]) using adjusted scores, and further to 78.3% with the optimized scores."},{"finding":"Validation accuracy for volatility direction rises from 53.8% (baseline) to 72.1% and 75.0% in subsequent models.","kind":"performance","quote":"Similarly, for volatility direction, accuracy increases from 53.8% in the baseline model to 72.1% and 75.0% in the corresponding models."},{"finding":"The expected average difference between original volatility and the hype‑adjusted volatility is approximately –0.0068 (–0.68%).","kind":"predictive-accuracy","quote":"Under a hype-adjusted probability measure of the selected date range, we calculate that the expected average difference between the original volatility and the adjusted one, based on the weights provided in the semiconductor dataset, is approximately −0.0068 (or −0.68%)."},{"finding":"Overall prediction accuracy improves by about 8% when using the hype‑adjusted sentiment scores.","kind":"predictive-accuracy","quote":"We note however, that despite this small number, the accuracy of the prediction was improved by 8%."},{"finding":"Accuracy increase of +8.3% for market return direction and +2.9% for volatility direction highlights refined precision from optimized sentiment modeling.","kind":"predictive-accuracy","quote":"The observed accuracy increase of +8.3% for market return direction and +2.9% for volatility direction highlights the refined precision achieved through optimized sentiment modeling."}],"reported_performance":null,"out_of_sample":{"method":"unclear","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":35,"citations":{"cited_by_in_corpus":1,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":1},"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":"2026-11-14T17:50:07Z"},"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/"}