Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2611.05778

Neural Score-Driven Volatility and Risk Forecasting

Qianli Zhao·Chao Wang·Minh-Ngoc Tran·Richard Gerlach
View original source ↗Read source document ↗Export metadata ↓

Abstract

We develop a score-driven volatility model in which no particular parametric innovation distribution needs to be specified. The proposed method relies on a lightweight neural network to represent the negative score of the innovation density. By integrating and normalizing the learned score and imposing location-scale identification, we recover a flexible innovation density that can accommodate asymmetry, including negative skewness commonly observed in financial returns. Two estimation procedures are considered. A two-step procedure estimates the neural score by score matching and subsequently fits the volatility recursion with the identified density held fixed. A one-step procedure jointly refines the neural and volatility parameters under the conditional likelihood. We illustrate the practical relevance of the proposed method using eight international equity indices, with a separate evaluation of the high-volatility first half of 2020. The results show that the neural score-driven models are competitive with Gaussian and Student-$t$ GAS benchmarks that require a prespecified parametric innovation distribution, with particularly strong performance during the COVID-19 period.

Explore related subjects

econ.EM

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Qianli Zhao, Chao Wang, Minh-Ngoc Tran, Richard Gerlach. 2026-11-04. Neural Score-Driven Volatility and Risk Forecasting. https://arxiv.org/abs/2611.05778

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Model fit vs prediction performance: contrasting fixed-coefficient models with mixture models

Mixture models such as mixed logit and latent class routinely achieve large gains in fit on panel data, yet several studies have found that these gains do not carry over to out-of-sample prediction. This paper explains why, building on the distinction between population-level and person-level prediction familiar from statistics. The panel log-likelihood of a mixture is exactly the cumulative score of forecasts that condition each choice on the person's earlier choices. Models are instead usually validated and applied with unconditional forecasts, which use only the estimated distribution of tastes and are scored by the cross-sectional log-likelihood. The difference between the two log-likelihoods, which we call choice consistency, comes from reproducing the dependence between a person's choices, and unconditional forecasts cannot use it. Decomposing the fit gain into choice consistency and heterogeneity, we show that when the assumed mixing distribution cannot reproduce this dependence exactly, pursuing choice consistency distorts the estimated heterogeneity, so that a mixture may predict single choices worse than a model without mixing. In two case studies, on revealed preference and stated choice data, choice consistency accounts for more than 90% of the fit gain for each of the fourteen panel mixtures. In our data, the part of the fit gain that helps to forecast single choices shrinks as people make more choices, and with many choices per person it becomes negative, leaving the mixture worse than a model without mixing. Mixtures forecast single choices clearly better only when the forecast can condition on the person's own earlier choices, which is often not possible in practice. For forecasting, the score of a panel mixture on single choices should therefore be reported alongside its fit, which requires only one further evaluation of the likelihood.

econ.EM↗
arXiv2026

Willingness-to-pay space and preference space: a unified treatment of equivalence, heterogeneity and interpretation

A growing number of choice modelling studies estimate their models "in willingness-to-pay (WTP) space", yet often treat this as a different model that fits better (or worse) than "preference space", and misread what its parameters represent. This note brings together and makes precise insights that have mostly been stated only in scattered places, organising them around a single condition for two specifications to be the same model. WTP space is a reparameterisation, and with non-random coefficients the two spaces give identical likelihoods, WTP estimates and standard errors. With deterministic heterogeneity in cost sensitivity, or non-linearity, WTP must still be computed as a ratio of partial derivatives, although a simple normalisation recovers a directly estimated WTP at a reference point. With random heterogeneity, latent class models are equivalent in the two spaces, while other mixtures are equivalent only when the implied distributions match, as with full-covariance lognormals, and independence is not a neutral assumption.

econ.EM↗
arXiv2026

Unobserved Consideration and Consumer Welfare

Can welfare effects of price change be identified when consumers' consideration sets are heterogeneous and unobserved? We establish that the welfare distribution is the same functional of structural choice probabilities under stable but limited consideration as under full consideration. The result permits unrestricted income-effects, preference heterogeneity, and dependence between preferences and consideration. Within this class, limited consideration admits a full-consideration representation preserving choice-probabilities and money-metric compensation: consideration/preferences can remain unidentified while demand and welfare are identified. When consideration sets contract as prices rise, the welfare distribution can be bounded using demand and a bound on choice-set attrition.

econ.EM↗