{"paper_id":"2402.17941","title":"Neural Networks for Portfolio-Level Risk Management: Portfolio Compression, Static Hedging, Counterparty Credit Risk Exposures and Impact on Capital Requirement","authors":["Vikranth Lokeshwar Dhandapani","Shashi Jain"],"paper_date":"2024-02-27","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2402.17941v1","page_url":"https://thequant.space/flowcharts/neural-networks-for-portfolio-level-risk-management-portfol/","scores":{"empirical_rigor":7.5,"math_complexity":8,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes an artificial neural network framework to compress a large portfolio of European options into a much smaller static hedge portfolio that replicates the target portfolio.","quote":"In this paper, we present an artificial neural network framework for portfolio compression of a large portfolio of European options with varying maturities (target portfolio) by a significantly smaller portfolio of European options with shorter or same maturity (compressed portfolio), which also represents a self-repli","summary":"In this paper, we present an artificial neural network framework for portfolio compression of a large portfolio of European options with varying maturities (target portfolio) by a significantly smaller portfolio of European options with shorter or same maturity (compressed portfolio), which also…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The underlying asset follows a Geometric Brownian Motion under a complete probability space.","type":"model","stated":"explicit","quote":"We assume a complete probability space (Ω, F, P), filtration Ft for an arbitrary time t ∈[0, T] and an adapted underlying asset process St, ∀t. The stochastic dynamics of the underlying asset are assumed to follow Geometric Brownian Motion (GBM), and therefore, St = S0 · exp[ (r −σ2/2) t + σ Zt ]."},{"assumption":"Each option in the portfolio is priced using the Black‑Scholes model.","type":"model","stated":"explicit","quote":"Let us consider the value of the target portfolio as V (t, St) at any arbitrary time t, where each constituent option is priced using Black‑Scholes pricing model introduced by [5]."},{"assumption":"The static hedge portfolio uses the same instrument type (vanilla European options) as the target portfolio.","type":"other","stated":"explicit","quote":"In both scenarios, the instrument type of static hedge portfolio constituents remains the same, resulting in a similar neural network structure to maintain the intended interpretation."}]},"datasets":{"status":"available","items":[{"name":"Test portfolio of 10,000 European options","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":"10,000 options (5,000 calls, 5,000 puts) with strikes 80‑120% moneyness and maturities 0.2","quote":"We assume a long target portfolio of 10,000 options with 5,000 call and 5,000 put options. The strikes of call and put options are selected equidistantly between 80% and 120% moneyness. The maturity of each target option is randomly selected from the discrete Uniform probability mass function Punif(U) : U →0.25, for ea"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"synthetic"},"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":"pricing-calibration","findings":[{"finding":"Exposure distributions and profiles (EE and PFE) of the target and compressed portfolios align closely across future risk horizons under both risk‑neutral and real‑world scenarios.","kind":"performance","quote":"We demonstrate through numerical examples that the Exposure distributions and Exposure profiles (Expected Exposure and Potential Future Exposure) of the target portfolio and compressed portfolio align closely across future risk horizons under risk-neutral and real-world scenarios."},{"finding":"The compressed portfolio replicates the target portfolio’s Delta and Gamma closely; Vega is lower when the compressed portfolio has a shorter maturity.","kind":"performance","quote":"We observed a close alignment of Delta and Gamma. In contrast, the compressed portfolio’s Vega was lower than the target portfolio’s Vega if the maturity of the compressed portfolio is lesser than the target portfolio."},{"finding":"Regulatory capital requirement under the standardized approach is reduced by 12%‑24% for the compressed portfolios compared with the original portfolios.","kind":"performance","quote":"Table 1 shows that the compressed portfolio’s capital is significantly reduced by 16.9%, 24.1%, 12%, 19.7% for the test portfolios in the same order."},{"finding":"Model error converges and neural‑network parameters (K and W) evolve smoothly across epochs at each time horizon.","kind":"theoretical","quote":"The model error at a specific time horizon is the absolute mean difference across the simulated paths between the target and compressed portfolio value."},{"finding":"The proposed methodology enables faster evaluation of exposure profiles at different risk horizons.","kind":"performance","quote":"The incremental advantage of the proposed methodology is faster evaluation of exposure profiles at different risk horizons."}],"reported_performance":null,"out_of_sample":{"method":"unclear","headline_out_of_sample":null}},"independent_evidence":{"status":"available","supports":[],"contradicts":[],"qualifies":[],"compared_with":23,"citations":{"cited_by_in_corpus":1,"by_independent_authors":1,"how_cited":{"used_as_baseline":0,"uses_method_or_data":1,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[{"paper":{"paper_id":"2311.12575","title":"Fast calculation of Counterparty Credit exposures and associated sensitivities using fourier series expansion","paper_date":"2023-11-21","authors":["Gijs Mast","Fang Fang","Xiaoyu Shen","Marnix Brands"],"empirical_rigor":7,"page_url":"https://thequant.space/flowcharts/fast-calculation-of-counterparty-credit-exposures-and-associ/"},"how":"uses_method_or_data","quote":null}],"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":"2026-11-14T18:14:08Z"},"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/"}