{"paper_id":"2407.19367","title":"Enhancing Black-Scholes Delta Hedging via Deep Learning","authors":["Chunhui Qiao","Xiangwei Wan"],"paper_date":"2024-07-28","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2407.19367v2","page_url":"https://thequant.space/flowcharts/enhancing-black-scholes-delta-hedging-via-deep-learning/","scores":{"empirical_rigor":8,"math_complexity":6.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a deep delta hedging framework that learns the residuals between the hedging function and the implied Black‑Scholes delta using neural networks.","quote":"This paper proposes a deep delta hedging framework for options, utilizing neural networks to learn the residuals between the hedging function and the implied Black‑Scholes delta.","summary":"This paper proposes a deep delta hedging framework for options, utilizing neural networks to learn the residuals between the hedging function and the implied Black-Scholes delta. This approach leverages the smoother properties of these residuals, enhancing deep learning performance. Utilizing ten…","strategy":"deep delta hedging learning residuals between hedging function and Black‑Scholes"},"assumptions":{"status":"available","items":[{"assumption":"Options are European‑style S&P 500 index options.","type":"market","stated":"implied","quote":"Our data comes from the OptionMetrics. We utilized ten years of data on S&P 500 index options, which are European-style options, covering the period from January 1, 2010, to December 31, 2019."},{"assumption":"Implied Black‑Scholes delta (practitioner delta) is used as the economic benchmark.","type":"model","stated":"implied","quote":"In practice, the implied Black‑Scholes delta, also called the practitioner Black‑Scholes delta, derived from market prices of options, is frequently used due to its incorporation of current market conditions."},{"assumption":"Data are filtered to exclude untraded quotes and those missing key fields.","type":"data","stated":"explicit","quote":"We followed the data filtering approach outlined by Hull and White (2017), excluding option quotes that were untraded or had missing information on bid prices, ask prices, implied volatility, δBS, θBS, vBS, or γBS."},{"assumption":"Options with time to maturity less than 14 days are removed.","type":"data","stated":"explicit","quote":"Additionally, we removed options with time to maturity less than 14 days, and chose call options with δBS in the ran…"}]},"datasets":{"status":"available","items":[{"name":"S&P 500 index option data","provider":"OptionMetrics","kind":"options","access":"commercial","period":"2010-01-01 to 2019-12-31","universe":"S&P 500 index options","quote":"Our data comes from the OptionMetrics. We utilized ten years of data on S&P 500 index options, which are European-style options, covering the period from January 1, 2010, to December 31, 2019.","known_id":"optionmetrics"}],"known":[{"id":"optionmetrics","name":"OptionMetrics IvyDB","provider":"OptionMetrics","kind":"options","access":"commercial","quote":"2 Data Our data comes from the OptionMetrics.","mentions":1,"used_in":1}],"sample":{"start":null,"end":null,"market":"S&P 500 index options","asset_class":"options","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"not_extracted","modeled":"unclear","headline_net_of_costs":null,"detail":null,"quote":null,"source":"none","rule":"R0","candidates":{"front_matter":"unclear","evidence_card":"unclear","evidence_card_quoted":false},"meaning":"the classifier was not confident"},"results":{"status":"available","evaluation_type":"pricing-calibration","findings":[{"finding":"Learning the residuals between the hedging function and the implied Black‑Scholes delta significantly improves hedging performance, often exceeding a 100% gain ratio improvement over directly learning the hedging function.","kind":"performance","quote":"Our empirical results demonstrate that this approach leads to a significant improvement in hedging performance compared to directly learning the hedging function, often exceeding 100%."},{"finding":"Adding input features improves the gain ratio more for put options than for call options, while market sentiment is less crucial when learning residuals.","kind":"performance","quote":"Adding input features when learning the residuals enhances hedging performance more for puts than calls, with market sentiment being less crucial."},{"finding":"Learning residuals with only three years of data achieves hedging performance comparable to directly learning the hedging function with ten years of data.","kind":"performance","quote":"learning the residuals with three years of data achieves hedging performance comparable to directly learning the hedging function with ten years of data."},{"finding":"For the model group using TTM and δBS as features, the residual‑learning model Fea2‑BS shows a 129% improvement for calls and a 101% improvement for puts in overall gain ratio.","kind":"performance","quote":"the model Fea2‑BS shows a 129% improvement in the overall gain ratio for calls and a 101% improvement for puts compared to the model Fea2."},{"finding":"Increasing the number of option‑characteristic and Greek input features further improves the gain ratio for residual‑learning models, especially for puts (up to 52% improvement).","kind":"performance","quote":"For puts, the improvement is more significant, reaching 52%."}],"reported_performance":{"strategy":"deep delta hedging learning residuals between hedging function and Black‑Scholes","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":24,"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":"2026-11-14T18:05:20Z"},"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/"}