Python code on asset centrality, crowded trades and impact on sector rotation
The code develops on the paper 'Crowded Trades: Implications for Sector Rotation and Factor Timing' by William Kinlaw, Mark Kritzman, and David Turkington in the 'Journal of Portfolio Management'.
Asset centrality (AC) is a statistical measure associated with a set of similar securities that reflects their ability to influence the return variability of a larger universe of securities to which the original set belongs. Higher (lower) levels of volatility and connectedness makes for higher (lower) ACs. The AC is higher for relatively more volatile sectors/factors as well as sectors/factors that are associated or more connected to other sectors/factors. The mechanics are translated in two ways:
1>via large order imbalances that result from large quantities of dollars used to buy or sell a particular sector/factor and produce large price changes and therefore higher volatility;
2> via decreasing diversification within sectors/factors as investors buy or sell the sector or factor as a unit as opposed to buying or selling individual constituents. The AC measure is a modification of principal components output into the absorption ratio (a measure of risk concentration) to determine the degree of influence on return variance present.
A relative valuation measure is added to distinguish scenarios where a significant change in AC is not accompanied by a change in fundamentals.