=== A Stock Market Recommender System ===

Christian Buchta, Robert Kofler, Andreas Mild and Thomas Reutterer
go back to:

COLLABORATIVE FILTERING virtual investmentclub stock market recommender system online recommendations

In recent years, recommendation systems have become an integral part of many online stores selling books e.g.,, dvds e.g.,, or music content e.g., A variety of techniques ranging from collaborative filtering to regression based methods are employed in such environments. We report from a novel application arena for such methods, namely an implementation of a stock market recommender system, called the virtual investment club VIC). As a basic premise this newly introduced system assumes that private investors' portfolio decisions are not solely based on classical decision criteria, but are also driven by latent preferences that are not covered by financial figures. VIC utilizes the time dependent nature of financial products and actively pushes new information to it's users. Recent trades are immediately recommended to users who hold similar portfolios. We are currently conducting experiments within an ongoing stock market competition and present first results of the comparative performance after a four month evaluation period.

©2006 by Christian Buchta, Robert Kofler, Andreas Mild, Thomas Reutterer