All knowledge is connected to other knowledge. The fun is making the connections.
-Arthur Aufderheide Paleopathologist "the mummy doctor"
I have been spending a significant amount of time studying the measurement of connectedness across markets and looking at network analysis. Much of finance is focused on the properties of a single market or a pair of markets, yet it may be better to think about markets as a complex network of relationships with each market representing a node and the connections are edges.
Some may view that network analysis is a just another way of thinking about correlation and they would not be wrong on an initial level, but thinking about markets as networks provides a deeper understanding of the connections between markets. Correlations between (x,y) and (y,x) are the same, but a network analysis can provide some insights on the direction of causality. It is extremely useful to understand whether a market shock has a spillover or contagion effect. Causality tests like those developed by Granger is one way to approach the problem but that analysis does not provide any visual insights of how a set of markets are connected.
While there has been significant work on networks and financial markets, it has not been mainstreamed and seems to be a backwater area for a small group of technical quants. Network analysis should be used as a fundamental tool for describing market relationships. From market networks, we can move to something deeper concerning the networks of trading agents. This requires a lot of data and computing power, but it would allow us to explore the behavior and actions of some agents on the overall market. This will get at the heart of market power, information advantages, and market flows.
I will be posting more on this very interesting area of finance.
No comments:
Post a Comment