---

name: volatility-models description: "Fit and troubleshoot PyFlux GARCH, EGARCH, EGARCH-in-mean, long-memory, and skew-t volatility models for heteroskedastic time series." disable-model-invocation: true metadata: disco-role: operating license: BSD 3-Clause

Volatility Models

Use this sub-skill for conditional volatility work on return series: data prep, model choice, leverage, in-mean effects, forecasting, intervals, and post-fit checks.

Route here

• Fit GARCH, EGARCH, EGARCHM, LMEGARCH, SEGARCH, SEGARCHM, or EGARCHMReg.
• Model volatility clustering, leverage, fat tails, skewness, or long memory in a univariate series.
• Use add_leverage() before fitting when the asymmetric news term is needed.
• Use fit, predict, predict_is, plot_fit, plot_predict, plot_predict_is, plot_sample, plot_ppc, plot_z, sample, and ppc on these volatility models.
• Validate with the bundled synthetic smoke section instead of the original Yahoo/FRED examples.

Reroute

• ARIMA, ARIMAX, or NNAR mean-model work: ../univariate-models/SKILL.md
• GAS / score-driven volatility work: ../gas-models/SKILL.md
• VAR or GPNARX work: ../multivariate-models/SKILL.md

Start fast

1. Convert prices to returns or log returns; if you do not have market data, use a local synthetic return series.
2. Pick the family:
   • GARCH for plain conditional variance
   • EGARCH for t-based volatility
   • EGARCHM for EGARCH-in-mean
   • LMEGARCH for long-memory volatility
   • SEGARCH / SEGARCHM for skew-t volatility
   • EGARCHMReg for regression-in-mean with exogenous regressors
3. Call add_leverage() before fit() when you need leverage terms.
4. After fitting, check summary(), plot_fit(), plot_z(), and then predict() / predict_is(); use sample() and ppc() only after BBVI or M-H.
5. Run the bundled smoke helper with --section volatility: smoke helper.

References

• API reference
• Workflows
• Troubleshooting
• Root families and inference for priors, fit methods, result objects, and posterior diagnostics.
• Root troubleshooting for install/build and legacy live-data failures shared across routes.