Variance-Guided Time-Series Modeling for Temporal Risk Detection // Fits balanced-panel autoregressive models with conditional heteroscedasticity for temporal risk detection. The main estimator combines autoregressive exogenous mean modeling with GARCH-X variance modeling, subject-specific baseline terms, shared population coefficients, and L1 penalization for high- dimensional covariates. The package returns conditional mean and variance estimates, coefficient summaries, simulations, and exceedance-based risk scores defined as estimated conditional threshold-exceedance probabilities. The implementation builds on the lasso of Tibshirani (1996) doi:10.1111/j.2517-6161.1996.tb02080.x, generalized autoregressive conditional heteroscedasticity of Bollerslev (1986) doi:10.1016/0304-4076(86)90063-1, and L1-regularized high-dimensional time-series modeling of Medeiros and Mendes (2016) doi:10.1016/j.jeconom.2015.10.011.