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

Nalan Bastürk; Peter C. Schotman; Hugo Schyns

doi:10.2139/ssrn.3871096

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

Nalan Bastürk, Peter C. Schotman, Hugo Schyns

Research output: Working paper / Preprint › Working paper

Abstract

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 language	English
Publisher	SSRN
DOIs	https://doi.org/10.2139/ssrn.3871096
Publication status	Published - 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"

Keywords

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

Access to Document

10.2139/ssrn.3871096

A Neural Network with Shared Dynamics for Multi-Step Prediction of Value-at-Risk and Volatility
Hugo Schyns (Speaker)
23 Jun 2021
Activity: Talk or presentation / Performance / Speeches › Talk or presentation - at conference › Academic

Cite this

@techreport{484fa120e1e343aab58fa91e890b46e1,

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

abstract = "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.",

keywords = "neural network, value-at-risk, volatility models, equity returns, risk management",

author = "Nalan Bast{\"u}rk and Schotman, {Peter C.} and Hugo Schyns",

year = "2021",

month = jun,

day = "21",

doi = "10.2139/ssrn.3871096",

language = "English",

publisher = "SSRN",

type = "WorkingPaper",

institution = "SSRN",

}

TY - UNPB

T1 - A Neural Network with Shared Dynamics for Multi-Step Prediction of Value-at-Risk and Volatility

AU - Bastürk, Nalan

AU - Schotman, Peter C.

AU - Schyns, Hugo

PY - 2021/6/21

Y1 - 2021/6/21

N2 - 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.

AB - 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.

KW - neural network

KW - value-at-risk

KW - volatility models

KW - equity returns

KW - risk management

U2 - 10.2139/ssrn.3871096

DO - 10.2139/ssrn.3871096

M3 - Working paper

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

PB - SSRN

ER -

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

Abstract

JEL classifications

Keywords

Access to Document

Activities

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

Cite this