In this paper we present a mathematical model for collaborative filtering implementation in stock market predictions. In popular literature collaborative filtering, also known as Wisdom of Crowds, assumes that group has a greater knowledge than the individual while each individual can improve group's performance by its specific information input. There are commercially available tools for collaborative stock market predictions and patent protected web-based software solutions. Mathematics that lies behind those algorithms is not disclosed in the literature, so the presented model and algorithmic implementation are the main contributions of this work.