{"paper_id":"2508.10778","title":"Dynamic Skewness in Stochastic Volatility Models: A Penalized Prior Approach","authors":["Bruno E. Holtz","Ricardo S. Ehlers","Adriano K. Suzuki","Francisco Louzada"],"paper_date":"2025-08-14","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2508.10778v1","page_url":"https://thequant.space/flowcharts/dynamic-skewness-in-stochastic-volatility-models-a-penalize/","scores":{"empirical_rigor":8.5,"math_complexity":9,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes dynamic skewness stochastic volatility models in the SMSN family (DynSSV‑SMSN) using priors that penalize model complexity.","quote":"In this paper, we propose dynamic skewness stochastic volatility models in the SMSN family (DynSSV‑SMSN), using priors that penalize model complexity.","summary":"Financial time series often exhibit skewness and heavy tails, making it essential to use models that incorporate these characteristics to ensure greater reliability in the results. Furthermore, allowing temporal variation in the skewness parameter can bring significant gains in the analysis of this…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The stochastic volatility model allows the skewness parameter to vary over time.","type":"model","stated":"explicit","quote":"allowing temporal variation in the skewness parameter can bring significant gains in the analysis of this type of series."},{"assumption":"Returns are heavy‑tailed and possibly skewed, justifying the use of SMSN distributions.","type":"statistical","stated":"explicit","quote":"The first approach is to expand the traditional SV model by incorporating possible heavy‑tailedness and skewness simultaneously, using a scale mixture of skew‑normal distributions."},{"assumption":"A weakly informative penalized complexity prior (PCP) is specified with hyperparameters (U, p) = (0.5, 0.5).","type":"model","stated":"explicit","quote":"For the PCP, the hyperparameter λ was specified using equation (10) with (U, p) = (0.5, 0.5), corresponding to a weakly informative prior."},{"assumption":"Parameter estimation is performed via Bayesian inference using Hamiltonian Monte Carlo (HMC) in RStan.","type":"statistical","stated":"explicit","quote":"Parameter estimation was carried out using the Hamiltonian Monte Carlo (HMC) method via the RStan package."},{"assumption":"Daily cryptocurrency returns are computed as mean‑corrected compounded percentage returns.","type":"data","stated":"explicit","quote":"Returns are computed as mean corrected compounded percentage returns, defined as yt = 100 × (log Pt −log Pt−1 −1/T"},{"assumption":"MCMC sampling uses 52,000 iterations with 50,000 warm‑up and retains the last 2,000 draws.","type":"statistical","stated":"explicit","quote":"For all models, we conducted the MCMC simulation for 52000 iterations, discarding the first 50000 as warm-up. The results found are based on the last 2000 iterations."}]},"datasets":{"status":"available","items":[{"name":"Daily returns of Bitcoin, Ethereum, and XRP","provider":"Yahoo Finance","kind":"other","access":"public","period":"January 7, 2017 to December 6, 2022","universe":"Three major cryptocurrencies (Bitcoin, Ethereum, XRP)","quote":"In this section, the proposed methodology is applied to the daily returns of three major cryptocurrencies, namely Bitcoin, Ethereum, and XRP. Data were obtained from Yahoo Finance (http://finance.yahoo.com) considering the period from January 7, 2017 to December 6, 2022.","known_id":"yahoo-finance"}],"known":[{"id":"yahoo-finance","name":"Yahoo Finance","provider":"Yahoo","kind":"market-prices","access":"public","quote":"Data were obtained from Yahoo Finance (http://finance.yahoo.com) considering the period from January 7, 2017 to December 6, 2022.","mentions":2,"used_in":2}],"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":"Penalized priors, especially the PCP, outperform classical priors in simulation across multiple scenarios.","kind":"statistical","quote":"Our simulation results show that penalized priors outperform the classical one in several scenarios, with the PCP yielding the best overall performance."},{"finding":"Dynamic models provide a better fit than static counterparts according to DIC, WAIC, and LOO‑CV.","kind":"performance","quote":"In all cases, lower values indicate better model fit. According to Table 8, the DynSSV‑t model exhibited the best performance across all three criteria."},{"finding":"Estimated skewness oscillates smoothly around zero, indicating alternating slight negative and positive asymmetry.","kind":"statistical","quote":"we find evidence of dynamic asymmetry that oscillates smoothly around zero over time, indicating that the conditional distribution of these returns tends to alternate between slightly negative and slightly positive asymmetry."},{"finding":"Models assuming normal errors perform poorly, yielding the highest information‑criterion values.","kind":"negative","quote":"models assuming normality yielded the highest values for the information criteria, further highlighting their poor fit under this assumption."}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":34,"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":true,"code_url":"https://github.com/holtz27/dynssv-smsn","note":null},"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":"2026-11-14T17:29:54Z"},"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/"}