• community detection
• community evolution
• Helsinki Stock Exchange
• investor network
• IPO
• over-expression
• SVN
• under-expression

This thesis analyses investor behaviour during the first two years after initial public offerings in Helsinki Stock Exchange. We use statistical validation method to construct investor networks based on co-occurrences of investors' trading actions. We find slightly different trading co-occurrences in the first and the second years. Also, we analyse the topological structure of the networks and we find that the number of small investor communities tends to decrease in the second year. In addition, we discover robust communities with a statistically signicant investor component that form across multiple securities and persist in time. Rich meta data of the analysed data set enables us to carry out the investor attribute over- and under-representation analysis. Finally, we detect statistically signicant difference in investor trading behaviour while analysing changes in monthly ownership positions and traded volume across multiple securities.