The basic proposition in this article is that there has been significant evidence (admittedly of a rather technical nature), demonstrated during recent market "difficulties", that there is considerable relative strength for the Russell 2000,the US micro-cap index, in comparison to the larger cap index the $S&P 500. After the tumultuous conditions across most asset classes over the last week or so, there are some interesting patterns emerging which point to the fact that the micro-cap stocks are behaving far more independently of macro risk than might be expected. In fact there is evidence going back to September 2010 that suggests that there is an asymmetry in the performance of IWM and SPY (I shall use the two ETF's IWM and SPY as surrogates for the cash indices).

One of the features of this asymmetry is that, during recent months, the beta value of IWM has been increasing when the overall market is performing well and has been decreasing when the overall equity markets are suffering from risk aversion. Based on a rolling 20 day linear regression the readings for beta (i.e. the gradient of the linear regression) were moving up towards a peak of about 1.8 during the strong upward impulse in prices during the latter part of 2010 and the first few weeks of 2011, and, during the recent price retreat, the beta value has declined sharply with a reading on Friday of just 1.14. This is not what one normally expects to see, as will be discussed below.

The chart below captures the relative price performance of three major US equity indices since September of 2010. The Russell 2000 has delivered a return of 25%, whereas the S&P 500 has returned 16% and the DJIA about 14%. Significantly the recent sell-off has been more severe for the large cap indices than for the 2000 smallest stocks that are traded on US exchanges and which are the constituents of IWM.



A second technique to illustrate the recent decline in the correlation or co-movement of IWM and SPY is to track the correlation between the daily price changes of each index via a linear regression. The regression is based on daily changes in each index and a 20 day sampling is rolled across the complete data set for the last two years (500 data points)

The chart below shows that…

Unlock the rest of this article with a 14 day trial

Already have an account?
Login here