I'm not exactly sure what to make of this, but I thought it might explain the relatively poor performance of the S&P500 & Russell 2000 pairs trading strategy in the last couple of years.
I looked at the 100 period standard deviation of each market and divided this by the previous day's closing level to translate standard deviation into some sort of percentage figure. I.e. a (not perfect) measure of volatility.
These rolling average show the volatility of the S&P 500 in blue compared to the S&P 500 in red. 0.15 = 15% etc. Interesting to note how volatility spikes tend to be accentuated on the Russel 2000.
The next chart shows the average difference in volatility between the S&P 500 and the Russell 2000. 1.00 = 100% i.e. twice as volatile.
Interesting to note how the cyclical this volatility difference is, with the average difference being 46% ie. the Russell 2000 is on average more volatile than the S&P 500 by half. However, since 2007 (the credit crunch era, the volatility difference has been below average has been below average. Perhaps a function of large cap bank stocks going nuts on the S&P 500, increasing its volatility relatively.
The difference seems to be creeping up again. Could this explain why the pairs trading strategy I've explored hasn't worked so well in recent years? The difference looks to be spiking once again.
One reader, Erik wondered if weighting the two markets makes a difference. To cut a long story short, it does, but not a huge amount. Weighting the two markets equally for the averages and for results actually reduces returns overall. It increases returns during the trend following pre 2000 era because the Russel 2000 gets a heavier weighting and reduces returns during the post 2000 era for the same reason.
It seems the Russel 2000 has rejected trend following stronger than the S&P 500 has taken up mean reversion.