{"paper_id":"2411.13599","title":"Can ChatGPT Overcome Behavioral Biases in the Financial Sector? Classify-and-Rethink: Multi-Step Zero-Shot Reasoning in the Gold Investment","authors":["Shuoling Liu","Gaoguo Jia","Yuhang Jiang","Liyuan Chen","Qiang Yang"],"paper_date":"2024-11-19","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2411.13599v2","page_url":"https://thequant.space/flowcharts/can-chatgpt-overcome-behavioral-biases-in-the-financial-sect/","scores":{"empirical_rigor":6,"math_complexity":4,"quadrant":"Street Traders"},"proposes":{"status":"available","statement":"The paper proposes a novel Classify‑And‑Rethink (CAR) method that uses Chain‑of‑Thought prompting to generate investment opinions and improve financial reasoning of LLMs.","quote":"We propose a novel method named Classify-And-Rethink (CAR) to generate investment opinion using Chain-of-Thought Prompting.","summary":"Large Language Models (LLMs) have achieved remarkable success recently, displaying exceptional capabilities in creating understandable and organized text. These LLMs have been utilized in diverse fields, such as clinical research, where domain-specific models like Med-Palm have achieved human-level…","strategy":"Classify-and-Rethink (CAR)"},"assumptions":{"status":"available","items":[{"assumption":"Buy/sell decisions are based on the average score of the latest five news articles compared to a threshold.","type":"execution","stated":"explicit","quote":"If the average score of the latest five news articles is below the threshold and the position is empty, we buy in with a full position; if the average score is above the threshold and the position is full, we sell out with an empty position."},{"assumption":"The model temperature is set to 0, max length 1024, presence penalty 1 to guarantee consistent results.","type":"model","stated":"explicit","quote":"The Temperature parameter of the API is set to 0, the maximum length is set to 1024, and the presence penalty is set to 1 to guarantee consistent results"},{"assumption":"Trading is executed at the daily closing price of the Shanghai Gold Exchange after news is reported at 15:30.","type":"execution","stated":"explicit","quote":"News reported after the Shanghai Gold Exchange’s daily closure at 15:30 is attributed to the next trading day."},{"assumption":"Large language models tend to maintain neutrality in emotional responses but can be misled by question framing.","type":"model","stated":"explicit","quote":"Compared to humans, large language models generally tend to maintain neutrality in their emotional responses. However, they are also susceptible to being misled by the way questions are asked."}]},"datasets":{"status":"available","items":[{"name":"Shanghai Gold Exchange spot gold index (Au9999.SGE) daily closing price","provider":"Wind","kind":"market-prices","access":"commercial","period":"2018-01-02 to 2023-06-13","universe":"Gold","quote":"The gold price used in the study is the daily closing price of the Shanghai Gold Exchange spot gold index(Au9999.SGE), which was obtained from the Wind database."},{"name":"Chinese gold news","provider":"http://www.dyhjw.com/","kind":"text-news","access":"commercial","period":"2018 onward","universe":"Gold‑related news","quote":"The gold news was collected from http://www.dyhjw.com/ and covers Chinese gold news from 2018 onwards."}],"known":[{"id":"wind","name":"Wind","provider":"Wind Information","kind":"market-prices","access":"commercial","quote":"The gold price used in the study is the daily closing price of the Shanghai Gold Exchange spot gold index(Au9999.SGE), which was obtained from the Wind database.","mentions":1,"used_in":1}],"sample":{"start":null,"end":null,"market":"gold","asset_class":"commodities","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"no","headline_net_of_costs":false,"detail":null,"quote":"Fig. 2 Asset value of different trading strategies without considering transaction costs.","source":"front_matter","rule":"R3","candidates":{"front_matter":"no","evidence_card":"no","evidence_card_quoted":true},"card_detail":null,"card_quote":"Fig. 2 Asset value of different trading strategies without considering transaction costs.","meaning":"a trading or execution result is reported without costs"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"CAR achieves the highest return (80.35%) and Sharpe ratio (1.071) among tested strategies.","kind":"performance","quote":"Classify+Rethink (CAR) 80.35% 1.071"},{"finding":"CAR outperforms other strategies both in returns and Sharpe ratio.","kind":"performance","quote":"Our CAR strategy outperforms other strategies both in returns and Sharpe ratio."},{"finding":"Score distribution from CAR is close to a normal distribution, better matching financial reality than other prompts.","kind":"statistical","quote":"The distribution of scores given by our CAR strategy is similar to the normal distribution, which is more aligned with the financial reality compared to the other two strategies."},{"finding":"Back‑testing shows CAR helps large models overcome the framing effect, leading to higher investment returns.","kind":"performance","quote":"the use of the CAR strategy helps large models to overcome the framing effect and score news, and the scoring distribution of news is more reasonable compared to unprocessed large models, leading to higher"}],"reported_performance":{"strategy":"Classify-and-Rethink (CAR)","sharpe":1.071,"sharpe_benchmark":0.811,"annual_return_pct":null,"total_return_pct":80.35,"max_drawdown_pct":null},"out_of_sample":{"method":"none","headline_out_of_sample":false}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":55,"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":"2026-11-14T17:53:39Z"},"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/"}