Réduction de la dimension d'un ensemble d'indicateurs techniques en vue de la prédiction de séries temporelles financières - Application à l'indice de marché BEL 20
(2000) ACSEG 2000 — Location: Paris (France) (14.December.2000)
Files
56-Réductiondeladimensiondunensembledindicateurstechniquesenvuedelaprédictiondesériestemporellesfinancières-ApplicationàlindicedemarchéBEL20.pdf
Restricted Access - Adobe PDF
- 637.68 KB
Details
- Authors
- Abstract
- Prediction of financial time series using artificial neural networks has been the subject of many publications, even if the predictability of financial series remains a subject of scientific debate in the financial literature. Facing this difficulty, analysts often consider a large number of exogenous indicators, which makes the fitting of neural networks extremely difficult. In this paper, we analyze how to aggregate a large number of indicators in a smaller number using -possibly nonlinear- projection methods. As the use of nonlinear projection tools involves the tuning of several coefficients, the core of this work is the development of a sound methodology to adjust these parameters, based on objective criteria. Several of these criteria are presented, and used in appropriated circumstances. Nonlinear projection methods are shown to be equivalent to the linear Principal Component Analysis when the rediction tool used on the new variables is linear. Furthermore, the computation of the nonlinear projection gives an objective way to evaluate the number of resulting indicators needed for the prediction. Finally, the advantages of nonlinear projection could be further exploited by using a subsequent nonlinear prediction model. It is also shown that the use of the determination coefficient is dangerous in the specific case of financial predictions. The methodology developed in the paper is validated on data from the BEL20 market index, using systematic cross-validation results. The application of this methodology on the BEL20 Market Index shows that comparable results are obtained when using a linear projection method or a non-linear one, when a subsequent linear prediction tool is used. The advantage of the methodology presented here is that it automatically evaluates the number of new variables that must be kept after projection, in order to keep the necessary and relevant information needed for the prediction. Finally, the advantages of a nonlinear projection could be further exploited by using a subsequent nonlinear prediction tool, even if the results would be more difficult to illustrate because of the inherent difficulties of nonlinear predictors.
- Affiliations
Citations
Lendasse, A., Lee, J., de Bodt, E., Wertz, V., & Verleysen, M. (2000). Réduction de la dimension d’un ensemble d’indicateurs techniques en vue de la prédiction de séries temporelles financières - Application à l’indice de marché BEL 20. ACSEG 2000 proceedings - Connectionist Approaches in Economics and Management Sciences, p. 155-175. https://hdl.handle.net/2078.5/253825