Exportar Publicação

Bravo, J. M. (2025). Ensemble Methods for Stock Market Prediction. In Rosa Meo · Fabrizio Silvestri (Ed.), Machine Learning and Principles and Practice of Knowledge Discovery in Databases. ECML PKDD 2023. Communications in Computer and Information Science. (pp. 430-448).: Springer Cham.

J. M. Bravo,  "Ensemble Methods for Stock Market Prediction", in Machine Learning and Principles and Practice of Knowledge Discovery in Databases. ECML PKDD 2023. Communications in Computer and Information Science, Rosa Meo · Fabrizio Silvestri, Ed., Springer Cham, 2025, vol. 2137, pp. 430-448

@incollection{bravo2025_1765700189423,
	author = "Bravo, J. M.",
	title = "Ensemble Methods for Stock Market Prediction",
	chapter = "",
	booktitle = "Machine Learning and Principles and Practice of Knowledge Discovery in Databases. ECML PKDD 2023. Communications in Computer and Information Science",
	year = "2025",
	volume = "2137",
	series = "",
	edition = "",
	pages = "430-430",
	publisher = "Springer Cham",
	address = "",
	url = "https://link.springer.com/chapter/10.1007/978-3-031-74643-7_31"
}

TY  - CHAP
TI  - Ensemble Methods for Stock Market Prediction
T2  - Machine Learning and Principles and Practice of Knowledge Discovery in Databases. ECML PKDD 2023. Communications in Computer and Information Science
VL  - 2137
AU  - Bravo, J. M.
PY  - 2025
SP  - 430-448
DO  - 10.1007/978-3-031-74643-7_31
UR  - https://link.springer.com/chapter/10.1007/978-3-031-74643-7_31
AB  - Forecasting stock market prices and returns is particularly difficult, given the nonlinearity, volatility, and complexity of the time series and the generally accepted, semi-strong form of market efficiency. Standard forecasting methods tend to follow a “winner-take-all” approach by which, for each series, a single believed to be the best method is chosen from a pool of competing times series, statistical learning, or machine learning methods. To cope with conceptual uncertainty and improve the accuracy of prediction in economics and finance time series forecasting, recent research investigated the use of dynamic model combinations. This paper investigates the performance of model combination approaches in financial time series forecasting. The set of methods includes a recent meta-learning strategy called Arbitrated Dynamic Ensemble (ADE), which is based on Arbitrating but dynamically combines heterogeneous learners by creating an embedded meta-learner for each base algorithm that specializes them across the time series. The findings show that: i) the ADE methodology presents a better average rank compared to widely used model combination approaches, including the original Arbitrating approach, Stacking for time series, Simple averaging, Fixed Share, or weighted adaptive combination of experts; ii) the ADE approach benefits from combining the base-learners as opposed to selecting the best forecasting model or using all experts; iii) the ADE method is sensitive to the aggregation mechanism.
ER  -