AdaVol: An Adaptive Recursive Volatility Prediction Method

Quasi-Maximum Likelihood (QML) procedures are theoretically appealing and widely used for statistical inference. While there extensive references on QML estimation in batch settings, it has attracted little attention streaming settings until recently. An investigation of the convergence properties procedure a general conditionally heteroscedastic time series model is conducted, classical optimization routines extended to framework large-scale problems. adaptive recursive routine GARCH models named AdaVol presented. The relies stochastic approximations combined with technique Variance Targeting Estimation (VTE). This method computationally efficient properties, while VTE alleviates some difficulties encountered by usual due lack convexity. Empirical results demonstrate favorable trade-off between AdaVol’s stability ability adapt time-varying estimates real-life data.