Statistical Evaluation of Deep Learning Models for Stock Return Forecasting

Firat Melih Yilmaz () and Engin Yildiztepe ()
Additional contact information
Firat Melih Yilmaz: Dokuz Eylul University
Engin Yildiztepe: Dokuz Eylul University

Computational Economics, 2024, vol. 63, issue 1, No 9, 244 pages

Abstract: Abstract Artificial intelligence applications, including algorithmic training, portfolio allocation, and stock return forecasting in the financial industry, are rapidly developing research fields. Particularly, the forecasting of stock return has received considerable attention. Selecting the most promising model to forecast stock return has always been the most prominent task in the literature. Several studies have reported the prediction ability of recurrent neural network (RNN) models, whereas only a few studies have rigorously evaluated the prediction performance of temporal convolution networks (TCNs) for stock return forecasting settings. Moreover, although most studies are focused on comparing the performance of deep learning models at a single horizon forecasting, the multi-horizon forecasting applications of stock return have been studied only limitedly. In this study, we aim to evaluate the forecasting performance of state-of-the-art deep learning models at multi-horizon forecasting paths in stock return forecasting tasks. Specifically, we focus on TCNs, RNNs, long short-term memory, and gated recurrent unit models in terms of forecasting performance and direction accuracy. Our evaluation framework based on the model confidence set shows which model(s) is (are) better, with statistical significance, for forecasting stock return at multi-horizon forecasting. The empirical results assert that the TCN model has the best out-of-sample forecasting performance at all forecasting horizons on three financial datasets compared to other models.

Keywords: Temporal convolutional networks; Recurrent neural networks; Model confidence set; Time series forecasting; Bayesian optimisation (search for similar items in EconPapers)
Date: 2024
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://link.springer.com/10.1007/s10614-022-10338-3 Abstract (text/html)
Access to the full text of the articles in this series is restricted.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:kap:compec:v:63:y:2024:i:1:d:10.1007_s10614-022-10338-3

Ordering information: This journal article can be ordered from
http://www.springer. ... ry/journal/10614/PS2

DOI: 10.1007/s10614-022-10338-3

Access Statistics for this article

Computational Economics is currently edited by Hans Amman

More articles in Computational Economics from Springer, Society for Computational Economics Contact information at EDIRC.
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().