Visualization of connectedness and networks - Moving beyond correlation

Everyone uses correlation to tell a story, but some cutting edge work takes this to another level when it is connected to concepts on networks and topology of mapping connections. Network research is on the cutting edge of understanding systemic risk and how capital and information flows across the globe. It can also be used to help visualize data. The idea that all markets will react together is a just a simplification that can lead to poor thinking about markets. A network approach can help show how markets interact.

I present some of the charts from the paper, "Evolution of world stock markets, correlation structure and correlation based graphs". It was published a few years ago but is still rather obscure. First. it provides some insights on how correlations change through time. Short-term correlation may go through swings based on selected shocks while correlations across longer horizons are more stable. However, looking at longer horizons and across calendar time provides insight on the structural change in connectedness of markets. 

Much of this work on correlations across time and over different horizons is known. What is less clear is how correlations can be decomposed to tell us about network connectedness. This provides a different way of looking at correlation. The use of such tools as Planar Maximally Filtered Graphs (PMFG) can provide significant insight on this "connectedness" of markets. Some of this may seen obvious for many equity markets, but it allows the data to speak for itself.

Correlations are "living" relationships that change with capital flows, opportunities, structural changes, and market events. Understanding the consecutiveness provides deeper insights on cross-asset behavior. I would like to say that I can see these relationships by just looking at a correlation matrix, but I will be lying. Visual displays of data can enhance our understanding of market behavior.