Fuzzy time series model based on red–black trees for stock index forecasting

Abstract

Forecasting data is still an extensively investigated area of research specially in stock markets. The subjectivity of the elements that influence the market oscillation is the main challenge that any forecasting model faces. In this context, existing fuzzy models have attempted to increase forecasting accuracy in financial markets over the years. Fuzzy returns of the phenomenon under investigation helps to mitigate the subjective part of the financial market, specially regarding the human feeling influence over it. Although there are several data structures that can help to define the proper clusters from the universe of discourse of a fuzzy model, this paper proposes a novel fuzzy model from which the universe of discourse is based on a red–black tree (RBT) data structure so as to increase the possibilities of obtaining better predictions. The RBT data structure is a binary search three data structure that promotes a better balance, which allows a better accuracy in the forecasting results. The proposed model is compared to well known fuzzy models in the literature showing better forecasting results.