An ideal investment is a company that has good long-term prospects and is currently undervalued.
The results of stock screeners often vary dramatically depending on market conditions, and don't give users a good way of organizing their search for new investments.
An ideal ranking system should be meaningful, not use empirically-derived weights/coefficients, and not have any arbitrarily chosen fudge factors.
In this article, I am going to explain a method I have been using to rank stocks, discuss the advantages and disadvantages of the method, and discuss how I use this method. There will be some algebra here, but I will also present my formula for anyone to use regardless of whether they follow my math (although I recommend understanding the math behind it).
First, a word about my investing philosophy. I have two basic criteria, both of which must be met, for investing in a company: (1) the company has good long-term prospects, and (2) the company is currently undervalued. If a company has good long-term prospects but is exorbitantly priced, I am not interested in buying, since I may not profit as a shareholder even if the company does well. On the other hand, if the company is undervalued but does not have good long-term prospects, then the company might go out of business or be bought out for a low price and not see any price appreciation, even if I was perfectly right about its undervalued-ness.
These criteria are simple enough, but there are plenty of investors who ignore them and will invest in companies with poor long-term prospects just because their valuation is so attractive, or investors who will invest in companies with excellent long-term prospects, but at exorbitant prices. Of course, the criteria are subjective, but investors are often aware when they violate these rules.
The advantage of following the two criteria is that they leave you some margin for error. If the company is undervalued but its prospects change, you are less likely to lose as much as if you invested in a company that already had poor prospects (since a change in prospects at that point generally means annihilation). On the other hand, if the company has good long-term prospects but suddenly becomes highly overvalued (in your estimation), well, then you can sell at a tidy profit. Following the rules will also give you a psychological advantage if you are invested in a company and the market moves against you. You are less likely to sell out at a loss if the company has good long-term prospects, since your patience will eventually be rewarded.
To summarize, my investing philosophy is essentially the same as Peter Lynch's, and not because I invented it independently. The disadvantage of my criteria, and perhaps all criteria, is that it can be difficult to find companies that are up to my standards. Thus, the need for a ranking system.
I prefer ranking systems over stock screeners, since the results of stock screens often vary dramatically depending on the current conditions of the overall market, whereas a company's ranking relative to other companies is in some ways more consistent, and in any case, why would you settle for a company that has somewhat good prospects and is fairly valued, when you can find a company that has the best prospects and the best value?
Unfortunately, stock screeners are generally the only free tools available, and I am not particularly interested in using other people's rating systems anyway, particularly if they don't describe their methodology. So here I am going to describe my own methodology for rankings, which might be helpful to other people who are interested in ranking.
My method is essentially a formula for determining the payback period of a given stock. To calculate this, I imagine buying an entire company and putting all of its earnings into my pocket. The time it would take me to recoup my investment is what I mean by payback period. By earnings, I mean EBITDA, even though EBITDA doesn't include things like necessary capital expenditures, taxes, etc., and obviously wouldn't all go into your pocket if you owned the business. However, EBITDA paints a more consistent measure of a company's earnings than net income, and even if it shortens the payback period, it will have a similar shortening effect for every company.
For the cost of buying an entire company, I use enterprise value, which is roughly given by: Enterprise value = market cap - cash + long-term debt. We can then write: EV = E1 + E2 + … + En , where Ei is the company's earnings in year i. Since the earnings are hopefully growing from year to year, we can re-write this as: EV = E + E(1+g) + E(1+g)^2 + … + E(1+g)^n, where E=E1 and g is the growth rate, which I will define shortly. Since we are looking for the payback period, we want to solve for n. After summing the geometric series, our formula for n then becomes:
n = log(1+g(EV/E))/log(1+g) - 1
For g, I want to include not just the company's growth rate, but the total return that a shareholder might expect to receive, so for this reason, I calculate g as compound annual revenue growth - compound annual shareholder dilution + current dividend yield. This is to get a better representation of the profits that accrue to shareholders, rather than just the owners of the company (who, after all, lose nothing by diluting their shareholders' stock, at least in the short term).
Historical revenue growth could easily be substituted with historical EBITDA growth, but since EBITDA can fluctuate more than revenue growth, I use revenue growth instead, with the implicit assumption that EBITDA is growing at the same rate as revenues. For historical growth rates, I use the information going back to the lesser of ten years or the oldest data that I have.
When I use "n" to rank stocks, I generally include the following fields along with the stock: Ticker symbol, company name, n, g_rev, g_share, g_div, and g. It's nice to be able to break the company down into these components to see which ones are contributing to the low n-value.
Some might wonder, why not simply rank the stocks by g? After all, this is the historical shareholder growth rate. The reason for this is that g is only an estimate of the company's long-term prospects, and doesn't take current value into account. In order for n to be low, however, the company must have good long-term prospects and be undervalued; i.e., it must meet both of the criteria that I discussed earlier. In the long term, companies with the highest g's should outperform everything else. However, in practice, it is difficult to extrapolate far into the future, and g can be highly subject to change, so it is desirable to find companies with both high g's and low n's.
There are several things that I like about this ranking method. First, it is intuitive, meaning that the n-value has meaning (the payback period), and doesn't use any empirically derived weighting coefficients or take analyst ratings into account, which are calculated using who-knows-what method. I am not fond of black-box methods, which have probably been back-tested over a certain period of history and may not necessarily be extrapolated forward.
Also note that the ranking method works for both dividend-paying and non-dividend paying companies, and isn't necessarily biased toward either. Dividend-paying companies are not unduly punished for having low growth rates, and companies that do not pay dividends are not unduly punished if they have substantial growth rates.
Furthermore, unlike DCF methods, the payback period method does not require any information about the overall market return, or require you to specify how long you wish to hold your investment (which can drastically alter your conclusions). The overall market return should be irrelevant to an individual company's ranking, and the holding period is simply a fudge factor, which I dislike, since they can be used to justify almost any conclusion. Finally, the method does not account for volatility at all, which I believe is mostly irrelevant to investing, since I am only concerned with the state of the underlying company and am willing to hold my investments for time periods long enough to make short-term volatility irrelevant.
At this point, you might wonder whether I have back-tested this formula and how my results fared. I have not back-tested the formula, and I absolutely do not recommend using it to mechanically invest. The reason for this is two-fold: First, in practice, there are a number of reasons a company might have misleading values for n and g, which I will discuss shortly. Second, many of the companies with the lowest n-values are terrible investments.
For example, almost every company with n<1 is a bad investment. These companies are mostly various Chinese industrial, pharmaceutical, and real estate companies, with numbers that look great on paper but have a dubious correspondence to reality. Other companies with n<1 were extremely cheap at the moment, but only because they had problems that hadn't yet manifested themselves on the balance sheets. I've found that the companies that interest me most generally have n-values between 1 and 4.
There are also a variety of practical reasons that companies might have misleading n-values. Some of these have to do with the difficulty of reliably processing missing or incomplete data, dealing with stock splits, etc. (this is a polite way of admitting that I am a bad programmer). Other times, there might be strange exceptions for holding companies or unusually large one-time dividend payments. Smaller companies can also have wildly fluctuating EBITDA values, which can temporarily make them appear to be extremely cheap (generally right before EBITDA plunges into the negative). Furthermore, very young companies often have explosive initial growth that is not sustainable or representative of potential future growth. Growth through acquisitions should also be distinguished from organic growth, since the former may not be as reliable for extrapolating g to the future.
For these reasons, the vast majority of companies with low n-values are actually not good investments. The ranking system is definitely not a substitute for good old-fashioned research. However, when viewed as a tool for finding new investments, I think that the ranking system is quite useful. The ranking system also works well with various filters to sift through the rubbish. Some extra filters I use, for example, are having g>10%, current revenues greater than $100M, companies that are 3 years or older, no precious metal mining companies, no for-profit education companies (which tend to appear high in the rankings), etc.
It's not terribly problematic for me if the majority of companies that show up on the list do not meet my standards, since I only keep about a half dozen companies in my portfolio at a time (hey, if I wanted diversification, I'd buy an index fund) and don't change them very frequently. The most useful thing about ranking is that it allows you to narrow your search for new investments down from over ten thousand stocks to maybe a hundred or so (depending on how far down the rankings you want to look), and gives you a way of organizing the search.
I hope that at the very least, this article has given you some food for thought, and of course, I welcome any questions or constructive criticisms regarding my method.
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.