In a recent Energy Letter, I discussed the data patterns and correlations that have historically provided reliable buy or sell signals for energy investments. This year such correlations have tipped us off to rare buying opportunities in natural gas as well as master limited partnerships. The returns, had you invested alongside our subscribers, would have been substantial.

I get a tremendous volume of energy-themed emails, and one recently included an intriguing claim. The idea is that when energy stocks make up no more than 7% of the S&P 500 by market capitalization, they outperform over the following three years.

It seems logical that whenever the energy sector’s weighting reaches some particularly low level, it may have become undervalued and subsequently outperforms. This is the kind of indicator I like. It makes sense, and the signal isn’t subject to interpretation.

But most importantly, is it true? I decided this one was worth investigating.

Getting the data was harder than I imagined. While there are lots of sources for the current sector makeup of the S&P 500, I couldn’t find any publicly available historical data on this. So I reached out to client support at S&P Global Market Intelligence, which, after some back and forth, provided 20 years of relevant data. Here is a graph of the energy sector’s historical weighting within the benchmark U.S. large-cap index:

The energy weighting has ranged from a low of 5.1% during the last week of 1999 to a high of 15.9% in July 2008, the month West Texas Intermediate topped $145/bbl. The average energy component weighting over the past 20 years was 9%.

Before analyzing the data, I did note a couple of things about this graphic. The first is that the all-time high in the weighting did precede a very sharp drop across the energy sector. In fact, between July 2008 and December 2008 the Energy Select Sector SPDR ETF (NYSE: XLE) slumped nearly 50%. In March 2011 the weighting climbed back above 13%, and over the next six months the XLE dropped some 20%. On the other hand, the weighting was at 14% in December 2008, at the outset of a five-year bull market that saw the XLE gain more than 120%.

Plotting the performance of the S&P 500’s energy components versus the weighting clearly shows these trends:

There’s a lot of information in that chart, but we can clearly see the two major bull markets in energy, in 2003-08, and 2009-14. Each ended in a steep sell-off. At the beginning of the 2003 bull market, the energy sector’s weighting was indeed below 7%. But it had been mostly below that level for four years. More on that below. As noted earlier, by the time the second bull market ended in mid-2008, energy’s weighting in the S&P 500 topped 15%.

But the 2009 bull market was quite different. The energy sector first bounced off the bottom when the sector weighting was still around 13% of the S&P 500. After an initial surge, the energy components traded sideways until late 2010, when they began a steep climb that would end in mid-2014. During that climb and the corrections that periodically interrupted, the energy weighting ranged from about 10% to 13%. When the plunge began in mid-2014, the energy weighting was just under 11%.

In a nutshell, there doesn’t seem to be a clear correlation between the energy share of the S&P 500 and its subsequent performance over a predictable period. From 2003 to 2008 the weighting trended up along with energy stocks, but then from 2009 to 2014 the weighting trended down while the share prices surged.

Now, circling back to the initial claim, when the energy share drops below 7%, does the sector outperform over the following three years? It may not be clear from looking at the graphic, so let’s break down the numbers.

The weighting briefly dropped below 7% in August of 1998, and moved decisively below that level in November 1998. The three-year performance of the S&P’s energy components from November 1998 through November 2001 was +3.4%. The S&P gained 11.6% over that time frame.

The weighting briefly popped above 7% in 2002 before falling back below for another two years. Had we bought into the energy sector in late July 2002, we would have realized a 129% gain by the S&P’s energy components over the next three years, versus a 47% increase in the S&P 500. So the signal seems to have worked on that occasion.

The problem, in terms of testing this theory, is that after energy’s S&P weighting once again again topped 7% in 2004, it only returned to that level in August 2015. We only have 13 months of data, but the S&P 500 is up 4% since, versus 8% for the S&P 500’s energy components.

The bottom line is that even though the data and graphics are interesting, they don’t provide strong evidence of a useful historical pattern. At least they do not meet my standards for something I would consider to be more reliable than not. There simply aren’t enough data points, especially considering that the first buy signal in 1998 would have been proved false.

So I won’t go so far as to call this a busted myth, but I can’t endorse this indicator (which was promoted by someone providing public advice on energy investing). I might even go so far as to say, knowing the factors behind the two energy bull markets of the past 20 years, that no indicator would have gotten investors in at the bottoms and and out at the tops. The turning points were dictated by a confluence of factors that can’t be summarized by a single, simple indicator.

This indicator is mostly a byproduct of wishful thinking and trying to project too much from insufficient data. I had to dig pretty deeply to get to the bottom of it, but sometimes avoiding a dead end is as important as blazing a new trail.  

(Follow Robert Rapier on Twitter, LinkedIn, or Facebook.)