A method to discover trend reversal patterns using behavioral data

Abstract

Cognitive biases often influence decision processes related to investment on stock markets. Mainly, this concerns complex problems with perception and understanding of surrounding financial and economic reality. This research was aimed at detecting cognitive biases in the data-driven manner. A few basic cognitive biases were examined: the Gambler's Fallacy, and the Hot Hand and Cold Hand effects. Detecting and modeling sequences leading to particular cognitive bias can significantly improve the trader's strategy. This paper presents a concept of a platform which can detect specific user behaviors. These are derived from the observation of technical analysis indicators, as well as the trader's own built-in indicators. Along with the standard functionalities of a stock market simulator, a few methods of data mining were applied: inductive decision trees, sequential association rules, clustering and visual exploration. © 2011 Springer-Verlag Berlin Heidelberg.