{"paper_id":"2506.06653","title":"Explaining Risks: Axiomatic Risk Attributions for Financial Models","authors":["Dangxing Chen"],"paper_date":"2025-06-07","paper_type":"theoretical","paper_url":"https://arxiv.org/pdf/2506.06653v1","page_url":"https://thequant.space/flowcharts/explaining-risks-axiomatic-risk-attributions-for-financial/","scores":{"empirical_rigor":3,"math_complexity":8.5,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes a risk attribution framework for machine learning models and extends the Shapley value to risk attribution problems.","quote":"We propose the framework of risk attribution problems for ML models.","summary":"In recent years, machine learning models have achieved great success at the expense of highly complex black-box structures. By using axiomatic attribution methods, we can fairly allocate the contributions of each feature, thus allowing us to interpret the model predictions. In high-risk sectors…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Baseline reference is zero and the explicand is greater than the baseline.","type":"model","stated":"explicit","quote":"Without loss of generality (WLOG), we assume x > x′ and x′ = 0."},{"assumption":"The target function f belongs to the set of real analytic functions.","type":"model","stated":"explicit","quote":"For simplicity, we assume F to be the set of real analytic functions and f ∈F."},{"assumption":"Error term ǫ is independent of the feature vector X.","type":"statistical","stated":"explicit","quote":"+ ǫ, where X is a m-dimensional random variable, ǫ is independent from X"},{"assumption":"Standard deviation is used as the risk measure in portfolio examples.","type":"market","stated":"explicit","quote":"Assume that the standard deviation is used to measure risk."},{"assumption":"Sub‑additivity holds for the risk measure, allowing diversification effects.","type":"statistical","stated":"explicit","quote":"We should expect to observe the impact of diversification when we calculate risk attributions."}]},"datasets":{"status":"available","items":[{"name":"Daily log returns of Tesla and Google (Alphabet)","provider":"Yahoo Finance","kind":"market-prices","access":"public","period":"2011 to 2023","universe":"Tesla and Google stocks","quote":"We apply a neural network to the daily log return of Tesla and Google (Alphabet) from 2011 to 2023 with industry factors as features.","known_id":"yahoo-finance"},{"name":"Industry factors","provider":"Kenneth French data library","kind":"macro","access":"public","period":null,"universe":"12 industry factors","quote":"We downloaded industry factors from the French’s website1. A total of 12 industry factors are used.","known_id":"french-library"},{"name":"Five selected stocks for CVaR optimization","provider":null,"kind":"market-prices","access":"unknown","period":"2023","universe":"Five stocks","quote":"For this example, we consider minimizing Conditional Value-at-Risk (CVaR) using α = 0.05 for the selected five stocks in 2023"}],"known":[{"id":"wrds","name":"WRDS","provider":"Wharton Research Data Services","kind":"other","access":"commercial","quote":"We collected stock prices and option data from Wharton Research Data Services in 2008.","mentions":1,"used_in":1},{"id":"yahoo-finance","name":"Yahoo Finance","provider":"Yahoo","kind":"market-prices","access":"public","quote":"The stock prices of Google and Tesla were collected from Yahoo Finance.","mentions":1,"used_in":1}],"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":"unclear","findings":[{"finding":"Optimizing the portfolio reduces CVaR from an average of about 0.038 to 0.019, demonstrating effective risk reduction.","kind":"performance","quote":"In the optimized portfolio, the CVaRα(Xc) is reduced to 0.019, which demonstrates the effectiveness of the optimization."},{"finding":"BShap risk attributions reflect diversification, with each component’s risk decreasing after optimization, especially Microsoft’s CVaR.","kind":"theoretical","quote":"Applying BShap (RAM), with the help of Proposition 3.2 for Sub-additivity, we are able to see the reduction of risks for each component by comparing the CVaR of X_i^c"},{"finding":"For the portfolio f(X1,X2)=X1+X2, RAM attributes zero risk to the risk‑free component X2, while BAM attributes its nominal value.","kind":"theoretical","quote":"By RAMs, we expect that there is no risk associated with X2 and therefore A2 = 0. By BAMs, there is an attribution from X2 with A2 = r."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":27,"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":"2026-11-14T17:35:59Z"},"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/"}