Quantile Convolutional Neural Networks for Value at Risk Forecasting

TL;DR

This paper introduces a novel quantile convolutional neural network approach for forecasting Value at Risk, enabling more accurate risk assessment by predicting specific quantiles of asset return distributions.

Contribution

It develops a modified CNN architecture capable of quantile forecasting for VaR, extending traditional CNNs to better capture distributional risk measures.

Findings

The model effectively learns from diverse asset price histories.

It produces fairly accurate VaR forecasts.

The approach enhances risk management tools.

Abstract

This article presents a new method for forecasting Value at Risk. Convolutional neural networks can do time series forecasting, since they can learn local patterns in time. A simple modification enables them to forecast not the mean, but arbitrary quantiles of the distribution, and thus allows them to be applied to VaR-forecasting. The proposed model can learn from the price history of different assets, and it seems to produce fairly accurate forecasts.