{"paper_id":"2511.12391","title":"Sharpening Shapley Allocation: from Basel 2.5 to FRTB","authors":["Marco Scaringi","Marco Bianchetti"],"paper_date":"2025-11-15","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2511.12391v3","page_url":"https://thequant.space/flowcharts/sharpening-shapley-allocation-from-basel-2.5-to-frtb/","scores":{"empirical_rigor":8.5,"math_complexity":7,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes novel practical solutions for negative risk allocations and a multi‑level Shapley allocation strategy that preserves additivity, together with a Monte Carlo implementation that remains feasible for many business units.","quote":"Moreover, we develop innovative solutions to address the challenge of negative risk allocations. Another contribution of our study is the development of a multi-level Shapley allocation strategy.","summary":"Risk allocation, the decomposition of a portfolio-wide risk measure into component contributions, is a fundamental problem in financial risk management due to the non-additive nature of risk measures, the layered organizational structures of financial institutions, and the range of possible…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Risk units follow a multivariate normal distribution and Value‑at‑Risk at 1% is used as the risk measure.","type":"model","stated":"explicit","quote":"To this scope we assume 5 risk units following a 5-dimensional multivariate normal distribution 𝑿, 𝑛= 5, 𝑿∼𝒩(0, 𝐶), 𝑋= ∑𝑋𝑖 𝑛 𝑖=1 ∼𝒩(0, 𝐶), (70) where 𝐶∈ℝ5×5 is the covariance matrix6, and 𝜌(𝑋) = VaR(𝑋, 1%).","support":"full"},{"assumption":"When the number of business units becomes large, the analytical Shapley formula is replaced by a Monte Carlo estimator that scales linearly with the number of units.","type":"execution","stated":"explicit","quote":"Specifically, we prove that the Shapley allocation strategy remains a feasible choice even in the challenging case of tens of business units, provided that, once a certain number of business units is exceeded, one switches from the analytical allocation (which scales exponentially with the number of business units) to the Monte Carlo allocation (which scales linearly with the number of business units),","support":"full"},{"assumption":"The computational cost depends on the effort required to compute the underlying risk measure ρ.","type":"other","stated":"explicit","quote":"M. Scaringi, M. Bianchetti Sharpening Shapley Allocation Page 20 of 46 As discussed in Sec. 2.3.4 and 3.1, the computational cost of the Shapley allocation strategy for 𝑛 risk units, measured in terms of calculations of the risk measure 𝜌, grows exponentially as 𝑛× 2𝑛−1 for Shapley analytical formulas and linearly as 𝑛× 𝑁𝑀𝐶2 ⁄ for Shapley Monte Carlo (which converges at rate 𝑂(1/√𝑁MC) using 𝑁𝑀𝐶2 ⁄ permutation","support":"full"},{"assumption":"The testing framework includes both simplified settings and realistic financial portfolios reflecting typical trading activities.","type":"data","stated":"explicit","quote":"Finally, we set up a specific testing framework, including both simplified settings, designed to highlight basic behaviours of allocation strategies of different risk measures, and realistic financial portfolios, which reflect the typical trading activities of a financial ins","support":"full"}]},"datasets":{"status":"available","items":[{"name":"Realistic financial portfolios under Basel 2.5 and FRTB","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"Our results have been demonstrated using a specific testing framework, including both simplified settings and realistic financial portfolios under Basel 2.5 and Fundamental Review of the Trading Book (FRTB) regulatory frameworks.","support":"full"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"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":"none","findings":[{"finding":"The Shapley allocation strategy offers the best compromise between simplicity, mathematical properties, risk representation and computational cost.","kind":"theoretical","quote":"Our results show that, in this context, the Shapley allocation strategy offers the best compromise between simplicity, mathematical properties, risk representation and computational cost.","support":"full"},{"finding":"Monte Carlo Shapley allocation scales linearly with the number of business units and becomes the only feasible choice already for n≈10‑14 depending on the number of Monte Carlo scenarios.","kind":"performance","quote":"The latter is still acceptable even in the challenging case of many business units, provided that an efficient Monte Carlo simulation is employed, which offers excellent scaling and convergence properties. Clearly, Shapley Monte Carlo becomes rapidly the only feasible choice, which, once adopted algorithm 1 in Sec. 3.1, scales as 𝑛× 𝑁𝑀𝐶2 ⁄ .","support":"partial"},{"finding":"The proposed methods successfully eliminate negative risk allocations while preserving the full‑allocation property.","kind":"negative","quote":"Moreover, we develop innovative solutions to address the challenge of negative risk allocations. To address this gap, we conduct a rigorous analysis of the theoretical conditions under which negative allocations may arise, and we propose some novel practical approaches that preserve the fundamental principles of Shapley allocation while ensuring non-negative outcomes.","support":"full"},{"finding":"The multi‑level Shapley allocation consistently preserves additivity across hierarchical levels of large financial institutions.","kind":"theoretical","quote":"Another contribution of our study is the development of a multi-level Shapley allocation strategy. Unlike other approaches, which typically focus on a single hierarchical level, or prevent marginal contributions from different hierarchical branches (see e.g. Li et al. (2016)), our framework systematically accounts for a consistent risk allocation which preserves the additivity property at multiple levels, a typical","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":22,"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":2,"extracted_at":"2026-11-14T19:46:02Z"},"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/"}