Bitte verwenden Sie diesen Link, um diese Publikation zu zitieren, oder auf sie als Internetquelle zu verweisen: https://hdl.handle.net/10419/317323 
Erscheinungsjahr: 
2019
Quellenangabe: 
[Journal:] Journal of Business Economics and Management (JBEM) [ISSN:] 2029-4433 [Volume:] 20 [Issue:] 1 [Year:] 2019 [Pages:] 168-191
Verlag: 
Vilnius Gediminas Technical University, Vilnius
Zusammenfassung: 
Forecasting the stock returns in the emerging markets is challenging due to their peculiar characteristics. These markets exhibit linear as well as nonlinear features and Conventional forecasting methods partially succeed in dealing with the nonlinear nature of stock returns. Contrarily, Artificial Neural Networks (ANN) is a flexible machine learning tool which caters both the linear and nonlinear markets. This paper investigates the forecasting ability of ANN by using Fama and French five-factor model. We construct ANN's based on the composite factors of the FF5F model to predict portfolio returns in two stages; in stage one, the study identifies the best-fit combination of training, testing, and validation along with the number of neurons full sample period. In stage two, the study uses this best combination to forecast the model under 48-months rolling window analysis. In-sample and out-sample comparisons, regression, and goodness of fit test and actual and predicted values of the stock returns of our ANN model reveal that the proposed model accurately predicts the one-month ahead returns. Our findings reinforce the investment concept that the markets compensate the high-risk portfolios more than mid and low beta portfolios and the methodology will significantly improve the return on investment of the investors.
Schlagwörter: 
artificial neural networks
forecasting
Fama and French 5 Factors CAPM
asset pricing models
stock markets
JEL: 
C18
C45
C63
G12
G17
Persistent Identifier der Erstveröffentlichung: 
Creative-Commons-Lizenz: 
cc-by Logo
Dokumentart: 
Article

Datei(en):
Datei
Größe
1.23 MB





Publikationen in EconStor sind urheberrechtlich geschützt.