{"paper_id":"2305.20067","title":"Modeling and evaluating conditional quantile dynamics in VaR forecasts","authors":["Fabrizio Cipollini","Giampiero M. Gallo","Alessandro Palandri"],"paper_date":"2023-05-31","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2305.20067v1","page_url":"https://thequant.space/flowcharts/modeling-and-evaluating-conditional-quantile-dynamics-in-var/","scores":{"empirical_rigor":7.5,"math_complexity":8,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper introduces new conditional quantile models: GARCQ within Direct‑Dynamics and three Indirect‑Dynamics specifications (QPI, Test Tracking and Multiplicative Tracking).","quote":"We introduce some new specifications: the GARCQ (inspired by the stationary GARCH) within the Direct-Dynamics specifications; and three new Indirect-Dynamics models, the QPI (inspired by the CAViaR of Engle and Manganelli (2004)) and the Violations Tracking approaches Test Tracking (TT) and Multiplicative Tracking (MT)","summary":"We focus on the time-varying modeling of VaR at a given coverage τ, assessing whether the quantiles of the distribution of the returns standardized by their conditional means and standard deviations exhibit predictable dynamics. Models are evaluated via simulation, determining the merits of the…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Actual quantiles follow a random walk: Et‑1 rqts = qt‑1.","type":"model","stated":"explicit","quote":"first, the actual quantiles follow a random walk, so that Et‑1 rqts = qt‑1; second, the bias bt in Equation (4) is zero (ct = qt‑1)."},{"assumption":"Bias term bt is zero (ct equals the previous quantile).","type":"model","stated":"explicit","quote":"second, the bias bt in Equation (4) is zero (ct = qt‑1)."},{"assumption":"The conditional distribution function F(pct‑1|ωt‑1) is strictly positively differentiable.","type":"statistical","stated":"explicit","quote":"assuming strictly positive differentiability of F(pct‑1|ωt‑1)"},{"assumption":"Logistic approximation to the indicator function introduces bias in the probability of exceeding the quantile.","type":"model","stated":"explicit","quote":"the logistic approximation introduces a bias in the probability of exceeding the quantile."},{"assumption":"Violations are tracked using an exponentially‑weighted estimator with decay λ ∈ (0,1).","type":"model","stated":"explicit","quote":"let us consider the exponentially‑weighted estimator ˆpt‑1 of the frequency with which given quantiles"}]},"datasets":{"status":"available","items":[{"name":"Fama‑French 25 value‑weighted portfolios","provider":"Fama‑French","kind":"market-prices","access":"public","period":null,"universe":"25 value‑weighted portfolios","quote":"The empirical application on the Fama‑French 25 value‑weighted portfolios with a moving forecast window shows substantial improvements..."},{"name":"Simulated returns with sinusoidal quantile pattern","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"As a check‑up analysis on this more familiar ground, we generate artificial returns with conditional variances that evolve according to the same sinusoidal pattern used for the quantiles."}],"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":"R4a","candidates":{"front_matter":"n/a","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"Incremental costs for the investor range from 10.44% to 487.12% for the dynamic GARCQ model.","card_quote":"the dynamic GARCQ (based on the realized quantiles from WHS99) performs worse than standard approaches, with incremental costs for the investor ranging from 10.44% to 487.12%.","meaning":"the paper reports no trading or execution result; the cost the extractor found is a model parameter"},"results":{"status":"available","evaluation_type":"forecast-accuracy","findings":[{"finding":"Forecasting conditional quantiles improves by keeping the predicted quantile unchanged unless the empirical violation frequency falls outside a data‑driven interval around τ.","kind":"predictive-accuracy","quote":"... shows substantial improvements in forecasting conditional quantiles by keeping the predicted quantile unchanged unless the empirical frequency of violations falls outside a data‑driven interval around τ."},{"finding":"MADτ rankings of constant and Indirect‑Dynamics models closely match RMSE rankings, whereas Direct‑Dynamics models diverge.","kind":"performance","quote":"the rankings induced by the MADτ on the constant and the Indirect‑Dynamics are very similar to those of the RMSE, while this is less true of Direct‑Dynamics."},{"finding":"Direct‑Dynamics specifications, especially standard historical simulations, perform poorly compared to constant‑quantile benchmarks.","kind":"negative","quote":"The results in Table 6 confirm the poor performance of the Direct Dynamics specifications. Standard historical simulations approaches, applied to the standardized returns, bring no improvements upon the constant‑quantile."},{"finding":"The Violations Tracking (TT) specification further damps parasitic oscillations in predicted quantiles.","kind":"theoretical","quote":"The TT specification further dampens such oscillations by updating the condit…"},{"finding":"When MADτ selects an Indirect‑Dynamics model, the infeasible RMSE agrees over 90% of the time; agreement drops to about 50% when MADτ prefers a Direct‑Dynamics model.","kind":"statistical","quote":"whenever an Indirect‑Dynamics model is selected by the MADτ, the rate of agreement of the infeasible RMSE is generally above 90%"}],"reported_performance":null,"out_of_sample":{"method":"walk-forward","headline_out_of_sample":null}},"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":"2026-11-14T18:26:28Z"},"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/"}