Centrality, influence and selective communication in an online investor network: Case Shareville

摘要

Online interactions form a growing share of the information exchange between investors, but existing research on online investor networks is still relatively scarce and focused on forex trading platforms. This thesis is the first study to observe and analyze user behavior in an online investor network focused on the stock market, specifically the Shareville community consisting of customers of the Nordic bank Nordnet. Its purpose is to investigate how activity and portfolio performance determine an individual investor's centrality in the Shareville network, whether investors influence each other's trading decisions, and whether the selective communication hypothesis, i.e. the tendency to discuss profitable trades more frequently than losses, holds in an environment where portfolio contents and returns are public information.This thesis consists of three studies regarding 1) factors determining the centrality of a user in the Shareville network, 2) the influence of central users on trades by their followers, and 3) selective communication in the network. In Study 1, regression analysis is used to determine how past performance and communication in the Shareville network are correlated with the number of followers a user has, and how users whose own followers are well-connected within the network differ from the baseline. In Study 2, buy-side trades made by users with followers are analyzed to determine how frequently the followers copy the trade by the end of the next trading day. In Study 3, returns from sell-side trades made by Shareville users are reconstructed by analyzing two years of trading data, and trade outcomes are combined with comments users make on their own trades. The commenting frequency on gains and losses is analyzed separately for Finnish and Swedish Shareville members.Results indicate that 1) historical portfolio performance and activity in the network are relevant but inadequate variables in explaining the number of followers a user has. Commenting on trades appears to be the most relevant form of activity in this regard, while e.g. comments in discussion groups or on individual instruments have less or no value. Results also provide 2) strong evidence of users copying trades of those users they are following. Finally, 3) the knowledge that portfolios and trades can be openly viewed by other users does not prevent users from practicing selective communication, so that trades resulting in gains are discussed more frequently than those resulting in losses. Statistically significant results on selective communication are obtained only for Swedish users, although for Finnish users the lack of statistical significance may result from a small sample size rather than complete absence of the effect.The results and discussion presented in this thesis will hopefully aid in further development of Nordnet's Shareville platform, inform future studies on the characteristics and behavior of retail stock market investors in online investor networks, and provide retail investors with a better understanding of their own behavior and that of others in the network - for example, how phenomena such as selective communication give a biased view of investor skill.