This thesis explores the development of a recommendation system for finan- cial instruments, specifically tailored for StockRepublic, a software company based in Stockholm, Sweden. StockRepublic offers a social platform for finan- cial institutions, enabling users to share and follow investment portfolios. The primary aim of this research is to devise a method for recommending new fi- nancial instruments to users based on their existing portfolios. The approach involves constructing a mathematical measure of similarity between instruments and developing an algorithm that leverages this similarity to suggest relevant additions to a user’s portfolio. Experimental results indicate that while the pro- posed similarity measure and algorithm function correctly, the overall similarity between instruments in typical portfolios is relatively low, resulting in poor end to end performance. This highlights the challenge of defining effective similarity metrics in the diverse field of data science.