Testing the generalised efficacy of technical analysis with bootstrapped aggregated regression trees

In this paper we examine the predictive power of the combined use of 23 known technical patterns and indicators, on 25 of the world’s most famous market indices, over the last decade. The system implemented for the combination of the above tools is bootstrapped aggregated regression trees, which is an ensemble nonparametric and nonlinear method and allows us to use numerical and categorical input variables simultaneously. Indications of inefficiencies are found, but their magnitude is not sufficient in order to characterise the aforementioned markets as weak form inefficient. In contrast, our overall conclusion suggests that technical analysis might marginally contribute in the interpretation of the manner that returns are evolved.