Debunked investment theories

We have here a collection of theories that seem attractive, or are popular, but which do not hold up under testing. Knowing what doesn’t work is as important as knowing what does, as making your screen more complicated than need be will hurt its maintainability, even if it does not hurt its performance.

Obviously, there may be something worth using here, but it will require refinement or further insight to be useful.

Buyback Programs – are they a fad?

Since the market crash in 2008 it seems that all big investment funds talk about when they make a large equity purchase is the fact that the company needs to add a buyback program. This is their way of eliminating share dilution and insuring that their piece of the pie becomes more valuable as time progresses. The question I have is, was this as common prior to the crash? How many companies are really doing buyback programs now compared to even ten years ago?

Here is an example of a backtest where we looked for all companies that decreased their shares outstanding over the past year. It beats the market, but only barely. On top of that, there isn’t a really meaningful increase in companies that are decreasing their outstanding shares in that time.

What about bigger companies?

Many large firms, like Berkshire Hathaway, focus on companies with larger market caps as they have been around longer and are less likely to fail in the near future. What happens if we put in a parameter that only grabs companies that are decreasing their shares outstanding and have a market cap of greater than $3 billion?

There we go! We see a noticeable change in the number of companies that match the screener as time goes on. However, it does look like our performance suffers and that we return almost equal what the S&P returns – and we do so at a higher standard deviation meaning our strategy takes on more risk for what is essentially the same return.

What about the extreme case?

Here we look for companies that retired 10% of their shares last year. ​​ That’s a lot of shares! ​​ As a convenience, we include the large cap and buying back shares (though not at the 10% level). ​​ 

As you can see, this is not a win. ​​ How about mixing both changes, to look only at large stocks with large buybacks?

That works better! ​​ So, are buybacks a fad? In short, no. Buybacks are just currently the best way to approach these large sum investments into these established (read: larger) companies. Are buybacks a harbinger of outperformance among large caps? ​​ Likely not. ​​ Notice that the bright green line (our “optimal” strategy) had a larger standard deviation. ​​ 

We are a victim of the law of small numbers. ​​ Smaller groups of stocks are more likely to have more extreme performance. ​​ At this level we can see that there’s essentially no alpha to be gained from buyback hunting.

OK, what is a death cross, and what do I mean by “it doesn’t work”? A death cross is a (very) vivid name for a short-term moving average slipping below a longer-term moving average.  Technical analysts love to sell when they see death crosses and buy when they see golden crosses.

Keep reading this article to see the application of death crosses in a screener and takeaways.

I just finished reading yet another book that advised me to avoid any companies cursed with the dreaded death cross. However, I’m the type of person to test out something myself rather than listen to what a book tells me.

Death Crosses Tested Quarterly

To warm up, let’s find a few companies that sport 50-day moving averages below their 200 days, found in the below image:

 Yowza! That’s a lot of stocks with 50-day moving averages below their 200-day moving averages… According to the technical analysts, these should all be dead stocks walking (should I have written this closer to Halloween?), but that seems like too many dead stocks to me.

Let’s see if this holds true. We’ll test it further, remember — green is good, and red is bad.

May 18-Aug 19th 2014,

That’s more green than I would’ve expected! Let’s try another quarter.

 Fed 15th to May 18th 2014

Better than I expected! Let’s try another one!

Nov 15th, 2013 to Feb 15th, 2014

Those zombies don’t look very dead to me! Heck, the zombies actually managed to beat the S&P 500 equal weighted performance one quarter. Testing the non-zombie stocks (those with 50-day averages above their 200-day moving averages) for the same three quarters yielded more stocks but lower performance. The “living” stocks never managed to outperform the S&P 500 average. But, you’ll say this was a silly test. We just tested three quarters. Maybe we should give it more time to work with… 

Death Crosses Tested Yearly

The following chart shows the living stocks (green and cyan) vs the zombies (purple) and SPY (the S&P 500 index fund, brown). As with the above screens, we’ve restricted ourselves to the S&P 1500 (smallcap, midcap and the S&P 500). We also rebalance weekly.

As you can see, the zombies were behind (slighty) in the period leading up to 2009, but they then outperformed from March 2009 onwards… The same data, different format, should make it easier to grok:

As we can see, even within the general trend, there is a lot of variability. What happens if we switch to an EMA (exponential moving average) rather than a simple moving average?

 It worked, yay! Wait a minute….. the purple line is the zombies — the stock we are not supposed to buy under any circumstances, and to sell as soon as possible. Maybe the backtest by time or log scale will show us something different.

Nope.

and nope. Maybe we’re beta blasting, and the zombies really really tanked when the market was terrible. Probably not — look at 2008, and how all the lines went down together.

Takeaways

It’s good to be cautious, but it’s better to be informed! If I had listened to that book, I would’ve ignored the opportunity to find these high-performing zombies and then would’ve believed that death crosses never have any positives.

If you’re the type of person that wants to see things for themselves, then Equities Lab is the best software for you! I was curious and decided to challenge the idea in that book and now I know death crosses can way outperform the living and the S&P 500. Find out for yourself using Equities Lab.

A Z-Score in statistics refers to how many standard deviations a particular data point is from the mean of the data. A Z-Score of 1 means the data point is one standard deviation from the mean. A Z-Score of 2 means two standard deviations. It’s useful when comparing data points from different sets of data.

The Altman Z-Score, on the other hand, is an unrelated financial model that was created by NYU business professor Edward Altman, and is used to predict a company’s likelihood of declaring bankruptcy. Below is the formula for the Altman Z-Score .

The How:

The Z-Score is calculated by weighing various business ratios and then adding them together. This number is then compared to a graded scale. It looks pretty complex, but it’s quite simple once you understand the different variables involved. I’ll break it down piece by piece.

A = working capital / total assets

B = retained earnings / total assets

C = earnings before interest and tax / total assets

D = market value of equity / total liabilities

E = sales / total assets

A Z-Score below 1.8 is an indicator of pending bankruptcy. A score of 1.8 to 3 indicates a company that might be headed to bankruptcy. Companies with a score over 3, are considered to be financially stable.

Let’s take a look at a backtest of the Z-Score formula in Equities Lab to see how accurate it has been. Below you can see the heat map and back tested chart of companies with an Altman Z-Score below 1.8, since 2000.

Well this is surprising. As you can see, of the many positions that we’ve held since 2000, the vast majority of them have performed well. Even more surprising, our portfolio of stocks that had poor Z-Scores, outperformed the market more than half the time. Perhaps the Altman Z-Score isn’t a good indicator of potential bankruptcy after all. Let’s continue our analysis by taking a look at companies that had Z-Scores between 3 and 1.8 for the same time period.

The surprises just keep coming.It seems like only a couple more companies with Z-Scores that indicate potential bankruptcy (1.8 – 3), performed better in comparison to firms with Z-Scores indicating impending bankruptcy. Additional, the firms that fit into our updated screener underperformed the companies that had poor Z-Scores early on, but then the two groups had identical performances after 2008. Furthermore, they still outperformed the market as a whole, more than half the time. Let’s see what the results were for firms with good Z-Scores before we completely disregard the test.

Surprisingly, or unsurprisingly, the backtest of firms with good Z-Scores didn’t perform well in comparison to our last two groups of firms. Even more unexpected, these firms didn’t outperform the rest of the market as much as the groups of firms with less than desireable Z-Scores. Perhaps the Altman Z-Score isn’t as prudent a tool for indicating firms that will go bankrupt. It’s probably a good idea to add additional variables in addition to the Altman Z-Score to draw out more accurate results.

High yield stocks are all the rage currently. Clients of mine are always asking for me to create screeners that search for the highest yielding dividend stocks in the market. This has resulted in me asking the question “how good are just straight high yield dividend stocks?”

To start this experiment off we are going to purchase all companies that have a dividend yield of greater than 0.1 (otherwise known as 10%).

backtest.jul_11_01_03_34


That went horribly. This strategy is incredibly volatile and you end up making less than the market in the past twenty years.  Let’s add some extra constraints on this strategy.

Here is the strategy if you make it so that it only returns companies that offer a yield greater than 0.1, and are capable of paying that dividend. We calculate this by making sure the EPS 1Y/Close is greater than the yield.

backtest.jul_11_01_05_52

Nope, that’s even worse. Maybe larger companies that meet these criteria are better investments.
This screen takes all of the above restraints and makes it so that only companies that pass those and companies that have a market cap over $1 billion are returned.  

backtest.jul_11_01_07_37

That’s not the ticket either. There may be some investments worth having in the very high yield space — but tread carefully.

Is Short Interest a Good Indicator?

In Equities Lab, if you subscribe to our professional level, you have the ability to access certain fields that aren’t available to our more basic membership levels. One such field is the short interest that is held in companies across the market, and it was this field that got me thinking. How good of an indicator is short interest in finding companies that are going to fall? At face value it seems reasonable that a company with a high short interest or a company whose short interest has drastically increased in the past few months would be more likely to go down, but let’s test it.

clip_image002

Well this chart makes absolutely no sense to me. It doesn’t give us an overwhelming movement in any direction. However, looking at it is like riding a rollercoaster and I’m starting to get a little woozy. This photo is of a simple screen that looks for short interest in respect to market cap and takes only the highest 99% percentile for highest ratio.

What about basing the screen off of the rank of the change in their short interest?

clip_image004

So it isn’t nearly as risky over the past 20 years, but it is still completely disproving the hypothesis we formed at the beginning of the article. This is a screen that is made up entirely out of companies where people are actively shorting their company at a higher rate than before. These companies should all fall, if other people’s trades are to be believed.

Maybe the time frame is wrong?

Maybe holding onto the short positions for a quarter is a bad idea. Let’s try a few different rebalance periods.

clip_image006

Here is the monthly rebalance. It had the exact opposite effect on the returns as I was expecting. It seems, that when it comes to the change in short interest, the price actually rises in the short term when companies have a large increase in short interest. Not something you’d want to invest in, but maybe not use this as a metric or rationale for why you short something.

image

On a little longer time scale – yearly rebalance, we at least don’t make more than the benchmark S&P 500. We instead make 100% over the past 20 years, so it still wouldn’t be a good short strategy no matter how you slice it.

This one is from ABG Analytics, a Quantitative investmnt site, and in their technical analysis section there is something known as the “Traditional Scoring Technique” This technique looks good on the surface and hits all of the basic technical analysis points, but let’s see how it fares.

clip_image001

Here is the formula we will be using. Each line is pretty self-explanatory, and the score itself goes from -6 to 6, but rarely does it go past five in either direction due to the RSI field parameters.

If you can’t tell from the formula, the higher the traditional technical score, the better. Therefore, we can assume that a high score is going to outperform the benchmarked S&P 500, and that it will outperform a lower score. We can also assume that if a company has a very bad traditional technical score, that it will underperform the market and possibly even lose money.

clip_image002

Here is an example of the formula used in an instance where our screen looked for companies that sported a Traditional Technical score greater than 3. It does beat the market by a bit, but it does so at a higher standard deviation – resulting in a Sharpe ratio of .13 compared to the benchmark’s .16. In this instance, though these companies were supposedly the best according to technical indicators. Now, the strategy did beat the market in that time period, even if only for a bit, so it does coincide with our assumptions.

clip_image003

That’s more along the lines of what I want to see when running tests. Here we isolated companies that had a traditional technical analysis score of less than -3 and backtested it over the past 22 years. This screen averaged .38% annually at a massive standard deviation.

clip_image004

When you get a screen that does so poorly it’s usually the result of not having a lot of results or a couple of the results bringing down the overall average. We have the exact opposite happening here. There are 170 matches, and the average number of positions held in your account at any one point in time is 320. That’s not a small pool.

Overall, the traditional technical score did pass the minor testing we did to it in this article. And though I wouldn’t use it as a green flag indicator to find something to buy, it seems to work fantastically as a red flag indicator for companies to stay away from.

editor.jul_11_03_41_23
backtest.jul_11_03_35_37
backtest.jul_11_03_41_28
backtest.jul_11_03_43_27

How important is income growth when it comes to finding a company to invest in?
In our screen we go through and look for companies that have a market cap above 500 million, a positive cash flow from continuing operating activities for the trailing 12 months for both three years ago and today, and the growth in the cash flow from continuing operating activities has grown over the past three years. Here are the results when we test this.
Yay! You make some money. You average 7.7% annually and you have a standard deviation of 5.68%. What happens if we put a cap on how much growth there is on income and let some companies lose money?
Here we isolate companies that exhibited income growth between -5% and 5%. Therefore, we are returning “Cash Cow” companies that have continuous income but little growth. Surprisingly, we returned more over the same time period with a lower standard deviation.  What if we isolate our results further to just companies that exhibit income growth between 0 and 5%?

We did worse… Shouldn’t cutting out the losers give us a better return? Our standard deviation went up, and we underperform that S&P 500 over the past four years. That wasn’t supposed to happen. Let’s take a look at companies that have an income growth between -5% and 0.

backtest.jul_11_03_44_46

That’s even more surprising. These companies had negative income growth over the past three years, yet it returned the highest return at 11.77% annually.

Final Thoughts:

Large, cash cow style companies don’t exhibit much growth in the way of income. However, many of these companies are incredibly stable and make good long term investments. Many of these companies are industry leaders in stable industries that can be considered “boring”.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.