A Neural Network with Shared Dynamics for Multi-Step Prediction of Value-at-Risk and Volatility

Research output: Working paper / PreprintWorking paper


We develop a LSTM neural network for the joint prediction of volatility, realized volatility and Value-at-Risk. Regularization by means of pooling the dynamic structure for the different outputs of the models is shown to be a powerful method for improving forecasts and smoothing VaR estimates. The method is applied to daily and high-frequency returns of the S&P500 index over a period of 25 years.
Original languageEnglish
Publication statusPublished - 21 Jun 2021

JEL classifications

  • g17 - Financial Forecasting and Simulation
  • c32 - "Multiple or Simultaneous Equation Models: Time-Series Models; Dynamic Quantile Regressions; Dynamic Treatment Effect Models"


  • neural network
  • value-at-risk
  • volatility models
  • equity returns
  • risk management

Cite this