Clustering Stocks with Self-organizing Maps: An application on Stocks Listed in BIST50 Index

The determination of characteristics of stocks are essential for an efficient portfolio management. In a well-diversified portfolio, the risk will be minimum. It is essential to know the characteristics of stocks for a better diversification. In this study, the stocks listed in BIST50 index are clustered using their standardized mean return and standard deviation of return. Closing price along 708 trading session is retrieved from Borsa Istanbul Datastore Department. For each stock, the last observation belongs to second trading session of 30/06/2015. Self-organizing maps which is a special kind of artificial neural networks are used as clustering technique. Also similarity matrix, scatter diagram, silhouette plots and time series plots for the selected stocks are drawn. Results indicate that self-organizing maps are successful at clustering and visualizing the stocks.