A0394
Title: Multivariate forecast for financial stock prices: A hybrid VAR-LSTM deep learning model
Authors: Diana Mendes - ISCTE-IUL (Portugal) [presenting]
Vivaldo Mendes - ISCTE-IUL (Portugal)
Nuno Ferreira - ISCTE-IUL (Portugal)
Abstract: The forecasting of stock price dynamics is a challenging task since these kinds of financial datasets are characterized by irregular fluctuations, nonlinear patterns, and high uncertainty changes. Deep neural network models, particularly the LSTM (Long Short Term Memory) algorithm, have been increasingly used by researchers and practitioners to analyze, trade, and predict financial time series, defining a new essential tool in several sectors' decision-making processes. The primary purpose focuses on a multivariate forecast of the U.S. stock index S\&P500, using Nasdaq, Dow Jones, and U.S. treasury bills for three months yields of the secondary market series, with daily frequency, between January 2018 and April 2023. With the support of a hybrid windowed VAR (Vector Auto Regressive) trend corrected by an LSTM recurrent neural network, we consistently obtain low forecast errors (around 4\%), even during the COVID-19 crisis. In addition, nonlinear Granger causality, based on transfer entropy, was tested between the periods with strong intervention by the Federal Bank, concluding that yield variation Granger causes the stock indices returns. In contrast, this causal relationship outside these periods was inverted, with the indices' returns causing yield variation.
