Period-aggregated transformer for learning latent seasonalities in long-horizon financial time series

Fluctuations in the financial market are influenced by various driving forces and numerous factors. Traditional financial research aims to identify the factors influencing stock prices, and existing works construct a common neural network learning framework that learns temporal dependency using a fi...

Full description

Saved in:
Bibliographic Details
Published in:PloS one Vol. 19; no. 8; p. e0308488
Main Authors: Tang, Zhenyang, Huang, Jinshui, Rinprasertmeechai, Denisa
Format: Journal Article
Language:English
Published: United States Public Library of Science 08.08.2024
Public Library of Science (PLoS)
Subjects:
Algorithms
Arbitrage
Computational linguistics
Computer and Information Sciences
Decomposition
Deep learning
Electric transformers
Engineering and Technology
Financial Management
Financial markets
Forecasts and trends
Humans
Information management
Investments - economics
Language processing
Latent period
Learning
Machine learning
Models, Economic
Natural language interfaces
Natural language processing
Neural networks
Neural Networks, Computer
Physical Sciences
Predictions
Principal components analysis
Research and Analysis Methods
Seasonal affective disorder
Seasonal variations
Seasons
Securities markets
Signal to noise ratio
Social Sciences
Stock prices
Task complexity
Time dependence
Time Factors
Time series
Transformers
Trends
Windows (intervals)
ISSN:1932-6203, 1932-6203
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Fluctuations in the financial market are influenced by various driving forces and numerous factors. Traditional financial research aims to identify the factors influencing stock prices, and existing works construct a common neural network learning framework that learns temporal dependency using a fixed time window of historical information, such as RNN and LSTM models. However, these models only consider the short-term and point-to-point relationships within stock series. The financial market is a complex and dynamic system with many unobservable temporal patterns. Therefore, we propose an adaptive period-aggregation model called the Latent Period-Aggregated Stock Transformer (LPAST). The model integrates a variational autoencoder (VAE) with a period-to-period attention mechanism for multistep prediction in the financial time series. Additionally, we introduce a self-correlation learning method and routing mechanism to handle complex multi-period aggregations and information distribution. Main contributions include proposing a novel period-aggregation representation scheme, introducing a new attention mechanism, and validating the model’s superiority in long-horizon prediction tasks. The LPAST model demonstrates its potential and effectiveness in financial market prediction, highlighting its relevance in financial research and predictive analytics.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
Competing Interests: NO authors have competing interests.
ISSN:1932-6203
1932-6203
DOI:10.1371/journal.pone.0308488