Pasar al contenido principal

Resultados relacionados

Resultados totales (Incluyendo duplicados): 2
Encontrada(s) 1 página(s)

e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publicación Artículo preliminar (WorkingPaper)

Information, Not Flexibility: Attributing Forecast Gains in Neural Stochastic Volatility

  • Guo, Hongfei
  • Marín, Juan Miguel
  • Veiga, Helena
When a neural volatility model beats a benchmark, is it the network or the data? We change one ingredient at a time within one Bayesian stochastic volatility framework, for three stock indices at one-, five-, and ten-day horizons. It is the data. Adding realised variance lowers the quasi-likelihood (QLIKE) loss by 2.8% to 13.4% in all nine cases and improves the one-day log score in every market. A neural network does not beat a linear function in one-day QLIKE with identical inputs, even inputs of the heterogeneous autoregressive (HAR) type. Combining the two neural models adds no stable gain. HAR and Realized GARCH have the lowest QLIKE in the nine-model comparison. With Student-t innovations, linear return feedback makes multi-step variance means infinite, while a bounded network does not. In practice, add realised information and choose the innovation distribution carefully, because it changes one-day tail risk rankings. Flexibility comes last., J.M. Marín acknowledges financial support from the Agencia Estatal de Investigación through project PID2020-115598RB-I00 and from Universidad Carlos III de Madrid through project 2024/00701/001. Helena Veiga acknowledges financial support from the Agencia Estatal de Investigación through projects PID2025-168576NB-I00 and PID2022-139614NB-C22, from the Fundação para a Ciência e a Tecnologia through project UIDB/00315/2020, and from Universidad Carlos III de Madrid through project 2024/00701/001.
e-Archivo. Repositorio Institucional de la Universidad Carlos III de Madrid
Publicación Artículo preliminar (WorkingPaper)

Testing Whether Volatility Model Gains Persist: A Prespecified Holdout in Tail Risk Forecasting

  • Guo, Hongfei
  • Marín, Juan Miguel
  • Veiga, Helena
Rankings of volatility models often change with the market, the evaluation period, and the forecast criterion, so a gain found in one setting may not carry over to another. We test this directly. Eight volatility models, four observation-driven and four stochastic volatility, are estimated for five equity indices by data cloning and compared on one-step-ahead forecasts. We assess joint value-at-risk and expected shortfall accuracy, calibration, predictive density accuracy, and forecast availability. In 2018–2023, nominally significant gains cluster in DAX and NIKKEI, with weaker evidence in FTSE. Neither holdout comparison is corroborated: the DAX estimate changes sign, and the NIKKEI estimate keeps its direction but is imprecise. Different criteria also favour different models. The results argue for validating a model in its intended market and for the criterion it is meant to serve, and they show how a prespecified holdout in a later period can test whether an advantage persists., J.M. Marín acknowledges financial support from the Agencia Estatal de Investigación through project PID2020-115598RB-I00 and from Universidad Carlos III de Madrid through project 2024/00701/001. Helena Veiga acknowledges financial support from the Agencia Estatal de Investigación through projects PID2025-168576NB-I00 and PID2022-139614NB-C22, from the Fundação para a Ciência e a Tecnologia through project UIDB/00315/2020, and from Universidad Carlos III de Madrid through project 2024/00701/001.