Comparing minimum spanning trees of the Italian stock market using returns and volumes

• An extended network analysis of 100 Italian stocks based on returns.
• Special care is taken regarding data sources, in particular corrections for dividends and capital operations.
• Four methods are adopted using symbolization rules and volumes.
• Reliability and centrality measures are used to compare methods and industry sectors.
• Petroleum and utilities sectors form clusters, insurances stocks are central.

We have built the network of the top 100 Italian quoted companies in the decade 2001–2011 using four different methods, comparing the resulting minimum spanning trees for methods and industry sectors. Our starting method is based on Person’s correlation of log-returns used by several other authors in the last decade. The second one is based on the correlation of symbolized log-returns, the third of log-returns and traded money and the fourth one uses a combination of log-returns with traded money. We show that some sectors correspond to the network’s clusters while others are scattered, in particular the trading and apparel sectors. We analyze the different graph’s measures for the four methods showing that the introduction of volumes induces larger distances and more homogeneous trees without big clusters.