On the Road to Nowhere
Daniel Drew, 12/15/2014
Oil has closed lower 9 out of the last 10 days. The price has crashed with an unfathomable steepness that has caught even the most experienced veterans off guard. The decline has cut down XLE by 25%, making it the worst performing sector for the quarter. It seems like everyone is trying to catch the bottom in either oil or XLE. Buy low, sell high, and win.
There's just one problem: sector rotation is a farce of a concept, and a simple reversion backtest shows unimpressive results.
The traditional idea of sector rotation is that certain sectors outperform the market at particular points in the business cycle, and other sectors underperform. Standard & Poor's even published a table so you can know for sure which sectors will be winners. Well thank God for the nice folks at S&P; - the same people who brought you the housing crash and who publicly admitted to making a $2 trillion calculation error when deciding whether to downgrade US government debt for the first time in 70 years. Yes, let's put our trust in their convenient sector rotation table on our quest to be the capitalist gods of tomorrow.
The most obvious problem with the sector rotation concept is that you have to accurately assess where we currently are in the business cycle. The track record for successfully doing this is pretty grim.
In January 2008, Merrill Lynch published a report that the United States was in a recession. Pretty decent call. In the ultimate irony, Lehman Bros disagreed with the report. Both Merrill Lynch and Lehman Bros would be on their knees later in the year, with Lehman Bros filing the largest bankruptcy in history and Merrill selling itself to Bank of America. Merrill's accurate prediction of the recession was not enough to save it from demise. Also ironic is that Martin Feldstein, the president of the National Bureau of Economic Research, denied Merrill's report and told CNBC, "I think we're not in a recession now." He did say that it could get worse, but he didn't actually raise the red flag. It wasn't until December 2008 - a full year after the recession had started - that the NBER said we were in a recession. By that time, 2.6 million Americans had already lost their jobs, making it the worst year in six decades.
Economist: There is no other job where stating the obvious is such an integral part of the job description. They are always 1 year too late.
Not only is the premise of sector rotation based on a fantasy, the execution of that premise is terrible. Even if you had your crystal ball or the keys to a magical DeLorean, sector rotation would be a waste of your time. If you actually knew at which point of the business cycle we were in, you could simply go long or short S&P; 500 futures with leverage. Forget sector rotation. It's nothing but complicated beta exposure.
Since no one knows where we are in the business cycle, for sector rotation to work, we would have to use some quantitative rules. As any backtester knows, you can't over-optimize your data, so you always start with the simplest rules first. I made the effort to conduct a basic test: Since 2009, look at the 9 sector ETFs that track the sectors in the S&P; 500. Every 6 months, buy the worst performing sector and short the best performing sector. If mean reversion in sectors is a real phenomenon, this strategy should lead to extraordinary returns. Unfortunately, you are only looking at gains of about 13%. That would be respectable if it were 1 year of returns, but it's for 5 years. So it's less than 3% per year - hardly a bankable effect and not something to bet the farm on.
So if you are salivating at XLE right now, and your finger is on the trigger, ask yourself how appealing 3% annual returns are. You could just buy a bond ETF and get similar results with less volatility.
The next time you hear someone talking about sector rotation, just visualize them on a carousel - going up and down on their high horse on the road to nowhere.
Update: It looks like Alex Gurvich of Seeking Alpha has a different way to calculate sector rotation. Instead of looking back at 6 month periods and holding for 6 months, he looks back 12 months and holds for 1 month. This method returns 49% from 2000 to 2012. Sounds great until you realize it's 4% per year. It's slightly better than the method I used, but nothing worth getting excited over. Such a small difference could just be a result of randomness.
XLE Gets Annihilated
Source: Yahoo Finance