I’ve been seeing a lot of articles recently that are very focused on the idea, that by purchasing the industry or sector that performed best over the past year you are giving you the best chance at making returns. I’m a little skeptical, but just as hopeful to see whether or not this is a potential new avenue to take in my portfolio.
Here is the base screen for what we want to do. This screener takes each sector and finds the total average change of close over the past year – making sure they have a market cap of at least $1 billion. It then ranks though sectors and selects the sector in position one.
Which sector did the best over the past year?
Much to my surprise, the answer is energy.
In the system, there are 232 companies classified under the Energy sector. Each sector contains a few different industries that all point back to the primary reason for the company existing.
The energy sector consists of the following –
- Oil & Gas – Integrated
- Oil & Gas – Services
- Oil & Gas – Midstream
- Oil & Gas – E&P
- Oil & Gas – Refining & Marketing
- Oil & Gas – Drilling
Yes, these are all oil & gas companies, but they all perform different tasks within the industry. If you’re thinking back to the oil crash in 2014, that crash mainly affected Drilling companies – though it did hit every industry within the sector to some degree.
Heading over to the Trading Rules tab, we can put the finishing touches on this strategy to make it trade annually.
The only change we need to make to what is already here is to rebalance this strategy annually on the first of the year. This will mean that we hold all of the companies within the top performing sector for an entire year before you even look at our portfolio again.
Good news, you beat the market…barely. Not only that, but it appears that the strategy left the building in late 2015 and never came back. Match that with a standard deviation that is nearly double the benchmarked S&P makes this philosophy incredibly worrisome.
Breaking the Sectors up Further
Maybe purchasing an entire sector which is made up of hundreds of companies is too broad a method. If we break it down further into individual industries, there is a higher chance for diversification.
To accomplish this change the “Sector” to “Industry”.
Like in the previous screener, our results belong to the energy sector, specifically Oil & Gas – E&P. But is this slight difference enough to change the fate of the screen?
Not so much. Yes, you did beat the S&P by quite a bit more than in the first backtest, but you also increased your standard deviation to 11% on a monthly basis. This increase in returns decreased your Sharpe Ratio to 0.09, whereas the first backtest sported a 0.11. Note: Neither of these Sharpe Ratios beat the S&P with a 0.173.
Worse yet, when you take a look at the performance of the individual companies that were taken on by this screener you realize just how loss heavy it is.
Usually, I like to start a screen with a 50/50 ratio of wins to losses and then build the screen from there without looking at specific details from either side. Remember, if you go into the losing companies and pick out the specific reasons they performed poorly, you run the risk of selection bias; You could end up future predicting and ruining the validity of your screen. To run effective screens you need to look at factors that affect the market as a whole rather than just individual companies.
Even when we break the Sectors down into industries, there are still quite a few too many companies to comfortably invest in as an individual. Of course, you can fix this by purchasing an ETF based on that sector, but that gives away a bit of the control. The way to mend this problem in the Equities Lab platform is to create a restrictions variable where we can create parameters to limit the companies that are found by the screen. You can put any number of items in this tab, but I’m only going to place two parameters.
I want the company to be large enough to have established a track record, and I want that track record over the past year to be positive by forcing companies to have at least made money over the previous year.
Now, there are two places we can input this new “restrictions” variable.
- You can place it after the first “where” in the editor. This might affect which industry is chosen as the strategy only takes the average change of close of companies within the industry that match the parameters placed after the first where.
- You can place it after the second “where” as well. This won’t affect which industry is chosen, but it does affect which companies from that industry appear in you screener.
Just to be double safe, and for the fun of it, I placed the restrictions variable in both places. Why not kill two birds with one stone?
Disappointingly, I only received two companies back from the original 75. It appears that very few Oil & Gas – E&P companies are worth $1 billion AND have a positive net income. Either way, they resulted in the highest average change of close over the past year so that we will go along with it.
When you run the backtest, it appears that this simple addition to the screen has made the returns extremely dull. But, even what seems to be a strategy that barely moves from the 0% line has a standard deviation of exactly double the market. How? Let’s take the log scale of the graph and get a better look at what’s going on with the returns of the test.
Though the strategy didn’t move much at the end of the day, it was a truly wild ride; wherein you lost 90% of your portfolio value at one point while bouncing all over the map. I don’t believe the restrictions are too tight, as it shouldn’t be that big of an ask to want a stable company that is making money. So, let’s try something different and see what happens if we change the rebalance period to happen more often. By doing this we are making our strategy buy and sell stocks more often which incurs trading costs, but we are also giving it a bit more variation.
We are going to leave the editor just as it is, and still look for the top performing industry over the past year. The difference is that we are going to create multiple points throughout the year for us to test. Going the most basic route, we will try a quarterly rebalance. At this point we are also going to add trading costs of 0.1% – not high, but we are attempting to only purchase fairly strong stable companies.
It worked a bit better, but still nothing to write home about. You still lose money when compared to the benchmark, but you do have steady returns over the past few years. This doesn’t really mean much, and honestly kind of just kills the strategy. I know there are people out there that have the different parameters when building an industry rotation strategy, and I’d love to see them in the comments so that I can try them out. Until then, I’m going to stay away from the industry rotation strategies in my portfolio.