Let's think about what that means. Technical analysts commonly refer to "divergences" among stocks at market turning points. That's one reason breadth measures are popular: they identify occasions in which stronger stocks are becoming differentiated from weaker ones. When we are making market lows, stronger stocks fail to follow the broad averages; when we are making market highs, weaker stocks fail to confirm the strength in the broad averages. The correlations among sectors provide a way of quantifying this differentiation among stocks.
If you click on the chart above, you can see that correlations have been highest around intermediate-term market bottoms and lowest around intermediate price peaks. During "risk off" periods, correlations rise considerably, as selloffs hit all sectors. The initial liftoffs from market bottoms find buying interest from longer timeframe participants and take most stocks off their lows. As rallies age, weaker stocks begin to lag and correlations fall.
If we go back to 2012 and create a median split of the data, we find that, when correlations are in the lowest half of their distribution, the next five days in SPY have averaged a flat performance. When correlations are in the highest half of their distribution, the next five days in SPY have averaged a healthy gain of +.68%. This pattern of subnormal returns for low correlation markets and superior returns for high correlation markets extends to 20 days out. Interestingly, correlations are at very low levels in the current market, suggesting reduced upside potential.
So we have all these measures that appear to have usefulness in anticipating forward price movement. How do we put them together to create truly evidence-based market indicators? That will be the subject of the next post in this series.
Further Reading: Intraday Sector Correlations
.
Posts
