{"paper_id":"2504.20088","title":"Deep Learning vs. Black-Scholes: Option Pricing Performance on Brazilian Petrobras Stocks","authors":["Joao Felipe Gueiros","Hemanth Chandravamsi","Steven H. Frankel"],"paper_date":"2025-04-25","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2504.20088v1","page_url":"https://thequant.space/flowcharts/deep-learning-vs.-black-scholes-option-pricing-performance/","scores":{"empirical_rigor":7,"math_complexity":6,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a deep residual neural network model, trained with a custom hybrid loss, to price European Petrobras options and compares it to Black‑Scholes.","quote":"A deep neural network was designed, inspired by residual network architectures commonly used in computer vision. The model was trained using a custom hybrid loss function that dynamically weights empirical market data and theoretical Black‑Scholes prices.","summary":"This paper explores the use of deep residual networks for pricing European options on Petrobras, one of the world’s largest oil and gas producers, and compares its performance with the Black-Scholes (BS) model. Using eight years of historical data from B3 (Brazilian Stock Exchange) collected via…","strategy":"deep residual network for option pricing"},"assumptions":{"status":"available","items":[{"assumption":"Stock prices follow a random walk and can be modeled using a Gaussian distribution.","type":"model","stated":"explicit","quote":"The BS equation is based on the idea that stock prices follow a random walk and can be modeled using a Gaussian distribution"},{"assumption":"Only European‑style contracts are considered; American options are excluded.","type":"market","stated":"explicit","quote":"the focus of this analysis was limited to Petrobras (PETR) European‑style contracts"},{"assumption":"Volatility is computed as a 21‑day rolling standard deviation of daily returns annualised by √252.","type":"data","stated":"explicit","quote":"we computed the volatility of PETR4 by taking the daily percentage changes in the stock price and applying a 21‑day rolling standard deviation. This rolling standard deviation was then multiplied by √252"},{"assumption":"If an exact date match for stock price, volatility or SELIC rate is missing, the closest preceding date is used.","type":"execution","stated":"explicit","quote":"If an exact date match was missing for any of these fields, we used the closest preceding date instead"},{"assumption":"Options with expiration between one and three months are retained; longer or shorter expirations are excluded.","type":"data","stated":"explicit","quote":"retaining only contracts set to expire within one to three months"}]},"datasets":{"status":"available","items":[{"name":"Petrobras options data (PETR) from B3","provider":"B3 (Brazilian Stock Exchange)","kind":"options","access":"public","period":"Nov 2016 to Jan 2025","universe":"Petrobras (PETR) European‑style contracts","quote":"The dataset contained information on all listed options, but the focus of this analysis was limited to Petrobras (PETR) European‑style contracts."},{"name":"PETR4 stock price and volatility","provider":null,"kind":"market-prices","access":"public","period":"Nov 2016 to Jan 2025","universe":"Petrobras common stock (PETR4)","quote":"To create a single dataset suitable for model training, we needed to combine the Petrobras option data, the PETR4 stock price (including its volatility), and the SELIC rate."},{"name":"SELIC rate","provider":null,"kind":"macro","access":"public","period":"Nov 2016 to Jan 2025","universe":"Brazilian risk‑free rate","quote":"To create a single dataset suitable for model training, we needed to combine the Petrobras option data, the PETR4 stock price (including its volatility), and the SELIC rate."}],"known":[],"sample":{"start":"2016-11","end":"2025-01","market":"Petrobras European options","asset_class":"options","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":"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":"pricing-calibration","findings":[{"finding":"The deep residual network reduces mean absolute error by 64.3 % for options priced between 3 and 19 BRL, covering 43.41 % of Petrobras option transactions on B3.","kind":"performance","quote":"For this price range, the model achieves a significant improvement of a 64.3% reduction in MAE across all months for around 43.41% of all Petrobras option transactions on B3."},{"finding":"The model maintains accuracy for longer expiration periods, contrary to Black‑Scholes which degrades over time.","kind":"predictive-accuracy","quote":"the deep learning model performed accurately for longer expiration periods."},{"finding":"The model overestimates option prices outside the 3‑19 BRL range.","kind":"negative","quote":"the deep learning model also has limitations, particularly a tendency to overestimate option prices outside the 3-19 BRL range."},{"finding":"The model successfully values contracts around 5 BRL that Black‑Scholes severely undervalues.","kind":"performance","quote":"it successfully values contracts priced around 5 BRL, such as PETRL36, PETRL410, PETRD397, and PETR409, whereas the Black‑Scholes model severely undervalues them."}],"reported_performance":{"strategy":"deep residual network for option pricing","sharpe":null,"sharpe_benchmark":null,"annual_return_pct":null,"total_return_pct":null,"max_drawdown_pct":null},"out_of_sample":{"method":"holdout","headline_out_of_sample":true}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":25,"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":"2026-11-14T17:40:07Z"},"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/"}