{"paper_id":"2508.18427","title":"Tracing Positional Bias in Financial Decision-Making: Mechanistic Insights from Qwen2.5","authors":["Fabrizio Dimino","Krati Saxena","Bhaskarjit Sarmah","Stefano Pasquali"],"paper_date":"2025-08-25","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2508.18427v2","page_url":"https://thequant.space/flowcharts/tracing-positional-bias-in-financial-decision-making-mechan/","scores":{"empirical_rigor":7.5,"math_complexity":4,"quadrant":"Street Traders"},"proposes":{"status":"available","statement":"The paper proposes the first unified framework and benchmark to detect, quantify, and mechanistically interpret positional bias in binary financial decisions using Qwen2.5 models.","quote":"We present the first unified framework and benchmark that not only detects and quantifies positional bias in binary financial decisions but also pinpoints its mechanistic origins within open-source Qwen2.5-instruct models (1.5B–14B).","summary":"The growing adoption of large language models (LLMs) in finance exposes high-stakes decision-making to subtle, underexamined positional biases. The complexity and opacity of modern model architectures compound this risk. We present the first unified framework and benchmark that not only detects and…","strategy":null},"assumptions":{"status":"none_found","items":[]},"datasets":{"status":"available","items":[{"name":"novel, finance-authentic dataset","provider":null,"kind":"benchmark","access":"unknown","period":null,"universe":null,"quote":"Our empirical analysis covers a novel, finance-authentic dataset revealing that positional bias is pervasive, scale-sensitive, and prone to resurfacing under nuanced prompt designs and investment scenarios, with recency and primacy effects revealing new vulnerabilities in risk-laden contexts."}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"none"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"unclear","findings":[{"finding":"Positional bias is pervasive, scale‑sensitive, and exhibits recency and primacy effects across financial prompts.","kind":"statistical","quote":"Our empirical analysis covers a novel, finance-authentic dataset revealing that positional bias is pervasive, scale-sensitive, and prone to resurfacing under nuanced prompt designs and investment scenarios, with recency and primacy effects revealing new vulnerabilities in risk-laden contexts."},{"finding":"Increasing model size from 1.5B to 14B reduces bias overall, but the Risk category remains stubbornly biased even at large scale.","kind":"statistical","quote":"stubbornly biased even at a large scale, suggesting domain-specific factors like volatility and exposure anchor the model to positional cues."},{"finding":"Prompt ordering independently influences bias magnitude; simple reordering causes significant shifts, especially for Innovation prompts.","kind":"statistical","quote":"Prompt ordering independently influences bias magnitude, with simple reordering causing significant shifts, especially in Innovation, emphasizing the need for deterministic prompt templates or randomized prompting with correction."},{"finding":"Prompt framing affects bias: conservative strong wording amplifies primacy bias, while moderate phrasing reduces bias.","kind":"statistical","quote":"Conservative, strong wording amplifies positional heuristics, likely reinforcing primacy bias, while Moderate phrasing promotes balanced evaluation and reduces bias."},{"finding":"Mechanistic analysis shows bias consistently originates in a stable set of mid‑to‑late transformer layers and a small recurrent group of attention heads acting as core bias engines.","kind":"other","quote":"positional bias consistently centers on a stable set of mid-to-late transformer layers and a small, recurrent group of attention heads that act as core “bias engines.”"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"available","supports":[],"contradicts":[],"qualifies":[],"compared_with":28,"citations":{"cited_by_in_corpus":5,"by_independent_authors":3,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":1,"mentions":2,"unclassified":2},"independent_papers":[{"paper":{"paper_id":"2609.02677","title":"Eliciting ESG Preferences for Reinforcement Learning-Based Portfolio Optimization","paper_date":"2026-09-02","authors":["Giovanni Dispoto","Marcello Restelli","Carmine Ventre"],"empirical_rigor":6,"page_url":"https://thequant.space/flowcharts/eliciting-esg-preferences-for-reinforcement-learning-based-portfolio/"},"how":"discusses_findings","quote":"Preliminary experimental results show that the medium-sized models considered here do not suffer from positional bias, unlike the smaller models in previous work [CITED]."}],"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":"2026-11-14T16:37:56Z"},"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/"}