Model Uncertainty and Forecast Combination in High‐Dimensional Multivariate Volatility Prediction

Abstract

In multivariate volatility prediction, identifying the optimal forecasting model is not always a feasible task. This is mainly due to the curse of dimensionality typically affecting multivariate volatility models. In practice only a subset of the potentially available models can be effectively estimated, after imposing severe constraints on the dynamic structure of the volatility process. It follows that in most applications the working forecasting model can be severely misspecified. This situation leaves scope for the application of forecast combination strategies as a tool for improving the predictive accuracy. The aim of the paper is to propose some alternative combination strategies and compare their performances in forecasting high-dimensional multivariate conditional covariance matrices for a portfolio of US stock returns. In particular, we will consider the combination of volatility predictions generated by multivariate GARCH models, based on daily returns, and dynamic models for realized covariance matrices, built from intra-daily returns. Copyright © 2015 John Wiley & Sons, Ltd

Can Speaker Augmentation Improve Multi-Speaker End-to-End TTS?

“…Metric for evaluating dialect confusion: Since dialect ID can be a challenging task even for native listeners, we evaluated confusion matrices of true vs. guessed accents. We computed Frobenius distance [37,38] between the confusion matrix for dialects of natural speech and those for each TTS system, based on the idea that if a confusion matrix for TTS is similar to the Table 3: Results: MOS and DMOS on a scale of 1-5 for seen (train) and unseen (dev and test) speakers. Synthesis was done using unseen texts.…”

Model-Preserving Sensitivity Analysis for Families of Gaussian Distributions

“…For this reason we recommend using the KL divergence in conjunction with another measure which takes into account the number of entries that have been varied. One such measure is the Frobenius norm, defined below, which has been recently used in econometrics and finance to quantify the distance between two covariance matrices (Amendola and Storti, 2015;Laurent et al, 2012).…”

Comparing the Performances of GARCH-type Models in Capturing Cryptocurrencies Volatility

“…However, given the large variety of volatility modelling options, model specification remains one of the main sources of uncertainty we need to deal with (c.f. [12], [37], [2]). In this paper, in light of these source of uncertainty, we evaluate and compare the performance of different models, within the GARCH class, in estimating and forecasting the volatility of cryptocurrencies by means of a Model Confidence Set (MCS) procedure as proposed by [33].…”