Paper: arXiv 2508.10778

Authors: Bruno E. Holtz, Ricardo S. Ehlers, Adriano K. Suzuki, Francisco Louzada

Abstract

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 type of series. However, for more robustness, it is crucial to develop models that balance flexibility and parsimony. In this paper, we propose dynamic skewness stochastic volatility models in the SMSN family (DynSSV-SMSN), using priors that penalize model complexity. Parameter estimation was carried out using the Hamiltonian Monte Carlo (HMC) method via the \texttt{“RStan”} package. Simulation results demonstrated that penalizing priors present superior performance in several scenarios compared to the classical choices. In the empirical application to returns of cryptocurrencies, models with heavy tails and dynamic skewness provided a better fit to the data according to the DIC, WAIC, and LOO-CV information criteria.

Complexity vs Empirical Score

  • Math Complexity: 9.0/10
  • Empirical Rigor: 8.5/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced Bayesian statistics with Hamiltonian Monte Carlo, complex skew-normal/Student-t distributions, and penalized complexity priors, indicating high mathematical density. It also presents rigorous empirical validation using simulation studies, information criteria (DIC, WAIC, LOO-CV), and application to real cryptocurrency data, demonstrating strong implementation and backtesting readiness.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Dynamic skewness in stochastic volatility models<br>with penalized priors for robustness"] --> B["Methodology<br>Propose DynSSV-SMSN models<br>using penalized priors"]
  B --> C["Computation<br>Parameter estimation via HMC<br>using RStan package"]
  C --> D["Simulation<br>Test penalized vs. classical priors<br>across multiple scenarios"]
  C --> E["Empirical Analysis<br>Cryptocurrency returns data<br>heavy tails & dynamic skewness"]
  D --> F["Key Findings<br>Penalized priors: Superior performance<br>Empirical: Better fit via DIC, WAIC, LOO-CV"]
  E --> F

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes dynamic skewness stochastic volatility models in the SMSN family (DynSSV‑SMSN) using priors that penalize model complexity.

Source sentence
In this paper, we propose dynamic skewness stochastic volatility models in the SMSN family (DynSSV‑SMSN), using priors that penalize model complexity.

Assumes 6

  • The stochastic volatility model allows the skewness parameter to vary over time. model
    Source sentence
    allowing temporal variation in the skewness parameter can bring significant gains in the analysis of this type of series.
  • Returns are heavy‑tailed and possibly skewed, justifying the use of SMSN distributions. statistical
    Source sentence
    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.
  • A weakly informative penalized complexity prior (PCP) is specified with hyperparameters (U, p) = (0.5, 0.5). model
    Source sentence
    For the PCP, the hyperparameter λ was specified using equation (10) with (U, p) = (0.5, 0.5), corresponding to a weakly informative prior.
  • Parameter estimation is performed via Bayesian inference using Hamiltonian Monte Carlo (HMC) in RStan. statistical
    Source sentence
    Parameter estimation was carried out using the Hamiltonian Monte Carlo (HMC) method via the RStan package.
Show 2 more
  • Daily cryptocurrency returns are computed as mean‑corrected compounded percentage returns. data
    Source sentence
    Returns are computed as mean corrected compounded percentage returns, defined as yt = 100 × (log Pt −log Pt−1 −1/T
  • MCMC sampling uses 52,000 iterations with 50,000 warm‑up and retains the last 2,000 draws. statistical
    Source sentence
    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.

Data

  • Daily returns of Bitcoin, Ethereum, and XRP · Yahoo Finance other public
    January 7, 2017 to December 6, 2022 · Three major cryptocurrencies (Bitcoin, Ethereum, XRP)
    Source sentence
    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.

Recognised sources: Yahoo Finance

Costs

Not applicable The paper reports no trading or execution result.

Finds 4

  • Penalized priors, especially the PCP, outperform classical priors in simulation across multiple scenarios. statistical
    Source sentence
    Our simulation results show that penalized priors outperform the classical one in several scenarios, with the PCP yielding the best overall performance.
  • Dynamic models provide a better fit than static counterparts according to DIC, WAIC, and LOO‑CV. performance
    Source sentence
    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.
  • Estimated skewness oscillates smoothly around zero, indicating alternating slight negative and positive asymmetry. statistical
    Source sentence
    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.
  • Models assuming normal errors perform poorly, yielding the highest information‑criterion values. negative
    Source sentence
    models assuming normality yielded the highest values for the information criteria, further highlighting their poor fit under this assumption.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 34 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

https://github.com/holtz27/dynssv-smsn — the authors' own repository

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (49)

  1. Stochastic volatility models with leverage and heavy-tailed distributions: A Bayesian approach using scale mixtures (2011)
  2. Penalising model component complexity: A principled, practical approach to constructing priors (2017)
  3. Bayesian analysis of stochastic volatility models (1994)
  4. The stochastic volatility in mean model with time-varying parameters: An application to inflation modeling (2017)
  5. Robust Bayesian analysis of heavy-tailed stochastic volatility models using scale mixtures of normal distributions (2010)
  6. Achieving shrinkage in a time-varying parameter model framework (2019)
  7. Stochastic volatility models with skewness selection (2024)
  8. The No-U-Turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo (2014)
and 41 more
  1. Riemann manifold Langevin methods on stochastic volatility estimation (2017)
  2. Stochastic Volatility Models for Asset Returns with Leverage, Skewness and Heavy-Tails via Scale Mixture (2014)
  3. The time-varying asymmetry of exchange rate returns: A stochastic volatility–stochastic skewness model (2020)
  4. Fitting nonlinear time-series models with applications to stochastic variance models (1993)
  5. Bayesian estimation of a skew-student-t stochastic volatility model (2015)
  6. Integrated nested Laplace approximations for threshold stochastic volatility models (2021)
  7. Simulated likelihood inference for stochastic volatility models using continuous particle filtering (2014)
  8. Generalized autoregressive conditional heteroskedasticity (1986)
  9. Modelling financial time series (1986)
  10. Pricing foreign currency options with stochastic volatility (1990)
  11. Quasi-maximum likelihood estimation of stochastic volatility models (1994)
  12. Estimation methods for stochastic volatility models: a survey (2004)
  13. Skew scale mixtures of normal distributions: Properties and estimation (2011)
  14. Smoothly mixing regressions (2007)
  15. Penalised complexity priors for stationary autoregressive processes (2017)
  16. The horseshoe estimator for sparse signals (2010)
  17. Bayesian variable selection in linear regression (1988)
  18. Parsimony inducing priors for large scale state–space models (2022)
  19. Econometric analysis of realized volatility and its use in estimating stochastic volatility models (2002)
  20. The variance-gamma distribution: A review (2023) — link
  21. The Variance-Gamma Distribution: A Review (2025)
  22. Density estimation for statistics and data analysis (2018)
  23. Penalized complexity priors for the skewness parameter of power links (2024)
  24. Evaluating interval forecasts (1998)
  25. Techniques for verifying the accuracy of risk measurement models (1995)
  26. Strategic long-term financial risks: Single risk factors (2005)
  27. Evaluating the accuracy of sampling-based approaches to the calculation of posterior moments (1991)
  28. Time-varying expected returns, conditional skewness and Bitcoin return predictability (2024)
  29. Value at risk estimation under stochastic volatility models using adaptive PMCMC methods (2017)
  30. Modelling and Forecasting Macroeconomic Risk with Time Varying Skewness Stochastic Volatility Models (2023) — link
  31. Stochastic volatility: likelihood inference and comparison with ARCH models (1998)
  32. Ancillarity-sufficiency interweaving strategy (ASIS) for boosting MCMC estimation of stochastic volatility models (2014)
  33. Informational efficiency of Bitcoin—An extension (2018)
  34. Analysis of the bitcoin stock market indexes using comparative study of two models SV with MCMC algorithm (2021)
  35. Forecasting the Risk of Cryptocurrencies: Comparison and Combination of GARCH and Stochastic Volatility Models (2024)
  36. Modelling and Predicting the Conditional Variance of Bitcoin Daily Returns: Comparsion of Markov Switching GARCH and SV Models — → 2401.03393
  37. Bitcoin and Ethereum Volatility Forecasting with GARCH Models (2024)
  38. Volatility dynamics of cryptocurrencies: a comparative analysis using GARCH-family models (2025)
  39. GJR-GARCH volatility modeling under NIG and ANN for predicting top cryptocurrencies (2021)
  40. Cryptocurrency risk management using L 'e vy processes and time-varying volatility (2025)
  41. Forecasting the volatility of the cryptocurrency market by GARCH and Stochastic Volatility (2021)

1 of 49 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.