One of the more predictable algorithmic stock market trades in recent years has been the Russell Reconstitution. Eminently predictable, investors have a darn good idea which stock will be added and deleted from the various Russell indices each year, and, better yet, they also knew when it would happen.
Why? Because the Richard Russell company has a published approach based on market capitalization, and a published calendar for when it will announce likely additions & deletions, and then when it will actually change the indices. So, it creates some fairly straightforward trades. Among others, for example, you could go long likely additions and short likely deletions, forming a kind of paired basket trade around the Russell reconstitution. Granted, you have to put the trade on before the Russell folks make the announcement, but knowing market capitalization that isn’t very hard to do.
While the trade has been reasonably successful for some time, it is starting to get much tougher. As a Smith Barney strategy paper points out, some of the problem has to do with hedge funds. Rather than guessing how and when index funds will slosh capital into and out of likely Russell additions or deletions, hedge funds now run so much money that they’re busily trying to outguess each other’s moves, the heck with the index funds. As a result:
…it almost looks as if 2004 was the mirror image of what was “supposed” to happen based on historical averages. In our view, however, what is really happening is that hedge funds are beginning to anticipate each other (as opposed to the index funds), and hence market moves are appearing earlier. If we “shift” the 2004 results two days forward (after already correcting for the June 25th rebalance date), the trends right around the rebalance date again appear to match up. be much more mild.
In a sense, it is an infinitely recursive version of Keyne’s famous analogy about picking stocks: The stock market is like a beauty contest where you don’t pick the person you think the most beautiful. Instead, you pick the person you think most likely that others will select. The trouble is, of course, this lends itself to infinite orders of recursion — Picking the person that you think that others will think that you will think that …
Related posts: