What We Can Learn From Sector Correlations in the Stock Market

Above you can see a proprietary index of stock market correlation that I have been working on in recent weeks.  The index looks across multiple stock sectors and multiple time frames and tracks the correlation of daily price moves.  In my modeling, correlation has been one of the most reliable predictors of prospective price moves in SPY over a several week horizon.  Indeed, going back to 2012, when correlation has been in lowest half of its distribution, the next 20 days in SPY have averaged a gain of .56%.  When correlation has been in the highest half of its distribution, the next 20 days in SPY have averaged a gain of 2.06%.

As market cycles age, fewer shares participate in the upward movement, as the weakest companies and industries first retrace their gains, followed by the stronger ones.  This causes the divergences we see among breadth measures and is directly measurable by correlation metrics that cover multiple market sectors.  As you can see from the chart, correlation has come down from its early August peak and is currently at levels associated with weaker next 20-day returns.  

My research has recently extended to intraday measures of correlation to see if they are predictive of shorter-term cycles in markets.  I also plan to study longer time frames and broad bull/bear market cycles.  I suspect there is also fruitful ground that could be covered by studying the correlation of movement of individual stocks within sectors as a way of anticipating sector rotation.  The interplay of correlation and volatility appears to provide useful metrics in tracking the trajectories of market cycles and the dynamics of momentum and reversal.

Further Reading:  Stock Market Buying and Selling Power
.