The Horseshoe Theory of Investing Styles
I cracked open my first book on investing 10 years ago. As I look back over the terrain I covered in my decade-long journey to learn investing, I see a landscape very different from the one I was told about in my early days. If I could give my younger self a map of the terrain, it would look very different from the one I was handed by those first few investing books.
The Map I Was Given
Like many investors, I started my journey with books such as The Intelligent Investor that teach “value investing”, by which they meant buying stocks with low P/E ratios. I later learned that the world’s greatest investor, Warren Buffet, would sometimes buy stocks with normal P/E ratios, but only if they were “wonderful companies”. There were a couple of successful investors, I heard, that would buy companies with P/E ratios that were a bit higher, but in general, only clowns would buy those stocks.
Of course, for the last decade, buying high P/E stocks has been a great strategy, but that is not why I now dislike this map of the investing landscape. I dislike it because it ignores an entire dimension along which investing approaches differ, and as a result, it misleadingly treats as different two styles that are actually the same in key respects.
My Own Map
The missing dimension in the map above is “quality”, by which I mean companies whose future earnings are especially predictable. These companies often have “moats” around them, as Buffett likes to say, protecting them from change and disruption.
When you add the dimension of quality (predictability), the landscape looks like this, forming the shape of a horseshoe rather than a line:
This new picture shows the commonality between the two extremes of the first picture — investing in high P/E stocks (hyper growth) or very low P/E stocks (deep value): these styles share the common trait of investing in situations with less predictable of outcomes.
This new picture also highlights the commonality between Growth at a Reasonable Price (GARP) and Buffett’s Wonderful Company at a Fair Price (WCAFP?): both approaches focus on quality companies with predictable growth going out at least several years into the future. Such companies can be valued more accurately using discounted cash flows and then purchased when offered at a discount to their intrinsic value. Whether those prices happen to correspond to above- or below-average P/E ratios (the bright red line in both pictures) is often not an important distinction here.
Left- vs Right-Skewed Portfolios
It is important to understand the difference between the styles at the top and bottom of the picture. Those at the top, which focus on quality companies, try to outperform by reducing bad outcomes. They focus on shrinking the “left tail” of the bell curve of potential results. Those at bottom, which buy into less predictable situations, try to outperform by generating more good outcomes. They focus on growing the “right tail” of the curve. The strategies at the top produce returns that are “left-skewed”, while the strategies at the bottom generate returns that are “right-skewed”, as shown here:
Note that both of these approaches work. Mathematically, either shrinking the left tail or growing the right tail will increase returns. However, these approaches are philosophically opposed to one another. A left-skewed portfolio looks for predictably above-average outcomes, which come from predictable (quality) companies, whereas a right-skewed portfolio looks for the potential for unusually good outcomes, and those are usually less certain.
Right-Skew, Style 1: Hyper Growth
I first encountered the right-skew philosophy when learning about David Gardner’s investing approach. He buys hyper growth companies with insane valuations. (Some of his companies don’t even have sensible P/E ratios because there is no “E” yet.)
Seeing this for the first time as a value investor, it was hard to make sense of it. The core of that difficulty, I believe, was the switch from a left-skew to a right-skew philosophy.
In the picture of skewed distributions above, even if the left- and right-skewed distributions have the same average outcome, they differ in that the bulk of the outcomes in the right-skewed distribution are on the lower end, below the average. This means that the probability of getting a worse-than-average (bad) outcome is higher in the right-skewed distribution even if the average result is the same in both cases.
When a distribution is right-skewed, more than 50% of the outcomes are below average (the median is less than the mean). That is why, when you look at a right-skewed portfolio, each stock, individually, looks like a bad pick: the most likely outcome is that it will underperform. Yet, the full portfolio can still outperform because the best outcomes are so good, that is, because the right tail extends so far out.
David Gardner emphasizes the right-skew philosophy in explaining his approach. To demonstrate how a large right tail can generate above average returns, he shows all the bad stock picks he has ever made and then how a single one of his big winners (and he has many) wipes out all of his losses.
Right-Skew, Style 2: Deep Value
What has become more clear to me in the last year is that investing in very low P/E stocks (deep value) is also a right-skew portfolio. The most likely outcome for most of these stocks is that they will underperform, but if they re-rate into even an average company, the resulting increase in their P/E ratio would imply a huge return for investors.
Here is a recent example. In the years before the pandemic shutdowns, GameStop often had a P/E between 5 and 7. It was priced like a company slowly going out of business because it was a company slowly going out of business. While it was making big profits, management was spending that money trying to find some way to extend the company’s life. If nothing changed, GameStop would eventually go bankrupt without returning much money to shareholders in the meantime. As above, the most likely outcome was a bad one (below-average or worse).
Even before the short-squeeze sent shares up 80x, GameStop showed a right-tail outcome. Multiple activist investors put pressure on the management team to start returning money to shareholders. As this materialized and became the most likely outcome, the share price increased 300% — an outcome from the far end of the right tail. A portfolio of three GameStop-like companies would have generated 33% returns for the year even if the other two companies went bankrupt, so even though each stock pick looked bad individually, the portfolio overall was set up for above-average returns.
Beyond this one example, low P/E stocks in general were hurt more during the crash in March 2020. They dropped more in price because they were riskier companies on average.
Historically, we see that deep value stocks are more sensitive to state of the economy. They get hurt worse during recessions but recover more afterward. This dependence on the state of the economy, which is hard to predict, makes the returns from a deep value portfolio also less predictable. The portfolio has outperformed historically, but with more booms and busts. It is a right-skewed portfolio.
Two Styles of Right-Skew
Those two styles of investing share the common property of generating right-skewed returns. They both have larger than usual winning outcomes, but their typical (median) outcome is worse than average. As I noted above, going into an individual trade with underperformance as the most likely outcome is hard for many investors to accept. It requires a different mentality.
Despite those commonalities, these two styles are very different from each other. Hyper-growth investors look for extremely high growth rates (above 50% per year) and extremely large total addressable markets. They ignore current valuations entirely. In contrast, deep value investors focus on the here-and-now: they look for companies that are earning lots of money relative to their current share price (low P/E ratios). Here, the distinction between growth and value is useful to separate the two styles.
One Style of Left-Skew
When we move to the left-skew style, on the other hand, the distinction between growth and value seems unnecessary at best and detrimental at worst. Of course, Warren Buffet told us this back in 1992:
In our opinion, the two approaches are joined at the hip: growth is always a component in the calculation of value, constituting a variable whose importance can range from negligible to enormous and whose impact can be negative as well as positive. [emphasis added]
It is only for quality (predictable) companies that near-term growth can be predicted with suitable accuracy, but when that is the case, Buffett tells us that growth is a component of value that we must consider to properly calculate intrinsic value. The bright red line between that separates high P/E (growth) from low P/E (value) stocks in our pictures above was not helping us identify undervalued quality companies — it was hurting our efforts.
The difficulty for those of us who learned to find undervalued stocks by looking for low P/E ratios is how to incorporate growth into a valuation.
Here is one simple approach: subtract the excess growth from the P/E. For example, if a company has a P/E of 21, but you expect it to grow 10% for the next five years while the average company grows around 5%, then you can subtract the excess (10 – 5 = 5) from the P/E, which reduces it to 16, a fair price for a good company.
If this calculation seems too cute to be correct, you are right, but the error is actually not where you might think: it is not in the resulting P/E but in the number of years. 16 is the correct P/E if the company continues to grow at 10% for around 5 years, but the precise number of years is probably not exactly 5. It is usually in the 4–6 year range, though, so if you believe strongly that the growth will last for at least 5 years before it starts to decline, then this is a safe estimate.
The other error in the calculation is that it actually holds for earnings that can be distributed as dividends, what Buffett calls “owners earnings” and others call “free cash flow”, and those are usually not all of reported earnings. If you want to be more conservative, you can shrink the excess growth by some amount — historically, 25% has worked well— to account for this. So now, the new P/E would be 21 – 0.75 * 5 = 17.25.
(The mathematically inclined reader can verify these calculations for themselves using the discounted cash flow definition of intrinsic value.)
These simple calculations should certainly not be taken as gospel, but hopefully this gets across the key idea that, for a predictable (high quality) company, whose future cash flows can be estimated with more-than-usual accuracy, a low-20s P/E can be a fair price if it is likely to grow 8–10% per year for 5+ years into the future, and a mid-20s P/E can be a fair price if it is likely to grow 12+% for 5+ years into the future.
There were plenty of times in the past decade when Google, for example, had a P/E in the mid-20s. While I did listen to Uncle Warren when he told me to buy wonderful companies at “fair prices”, I wish I knew back when I started learning 10 years ago that a mid-20s P/E could actually be a fair price for owning one of the world’s greatest companies. This simple calculation would have told me that.
