N-tuple patterns: A recent method to forecast market movements, with applications

Abstract

A new decision making strategy to forecast future market movements of stock prices conditioned to the past pattern will be presented, specifically patterns from one day up to seven, of a given index. The inspiration of such approach is given by the work of A. G. Malliaris, “N-tuple S&P patterns across decades, 1950-2011”. To start, in the first chapter, a review of the major financial techniques for forecasting market movements will be presented; the analysis will be done in a critical way, in order to understand strengths and weaknesses of such theories. In the second chapter the novel approach of Malliaris will be explained in details: he collected data from the S&P500 index for 60 years and, after they have been classified as Up or Down movement (with respect to the previous day), they have been analysed decade by decade. In the third chapter the same approach of Malliaris will be applied on the Italian index FTSE MIB. The whole work proves that certain conditional forecasts outperform the unconditional Random Walk model, demonstrating that prices do not move randomly but they follow a trend, based on the past.