Confidence Interval Construction for Multivariate time series using Long Short Term Memory Network

TL;DR

This paper introduces a new method for constructing confidence intervals for multivariate time series predictions using LSTM networks, employing novel bootstrap techniques and block length selection, validated on financial datasets.

Contribution

It presents a novel bootstrap-based procedure with innovative block length selection for confidence interval construction in multivariate time series using LSTM.

Findings

Effective confidence intervals constructed for financial indices

Comparison benchmarks demonstrate the superiority of proposed bootstrap methods

Method applicable to real-world multivariate financial data

Abstract

In this paper we propose a novel procedure to construct a confidence interval for multivariate time series predictions using long short term memory network. The construction uses a few novel block bootstrap techniques. We also propose an innovative block length selection procedure for each of these schemes. Two novel benchmarks help us to compare the construction of this confidence intervals by different bootstrap techniques. We illustrate the whole construction through S\&P $500$ and Dow Jones Index datasets.