I have an investing "paradox" for you. I'm curious to hear your thoughts. I start by dividing the value investing universe into two groups: 1) Ben Graham/Walter Schloss/net-net style and 2) Greenblatt/Magic Formula/Modern Buffett style. In very rough terms, category 1 suggests that you should buy low-P/E, low-P/B stocks ideally with low leverage. Category 2 suggests that you buy low-P/E, high-ROIC stocks ideally with low leverage. Both schools of thought basically agree that low P/E and low leverage are desirable. The main difference is that Graham looks for a low P/B ratio and Greenblatt looks for high ROIC.
Now consider the following exercise. Starting with the whole universe of stocks, first let’s eliminate stocks with high debt. To be concrete, let’s say we keep the 20% of stocks with the lowest net debt/equity. Second, lets further filter this group by only keeping the 20% of stocks with the lowest EV/EBIT (or P/E ratio if you prefer). At this point we have a group of relatively low-P/E, low debt companies. Now this is where the "paradox" begins. If I asked Graham whether he would like to invest in the low-P/B or high-P/B stocks in this group, he would obviously prefer the low-P/B stocks. Greenblatt on the other hand would say that he wants the highest ROIC stocks in the group, which is effectively saying that he would like the high P/B stocks in the group. So, they would pick from opposite ends of the spectrum.
The "paradox" is that both strategies are claimed to work well. But how can this be? They are choosing from opposite ends of the P/B spectrum.
Is there another way to resolve this? Have I over-simplified the Graham and Greenblatt schools of thought beyond recognition? Have I made a logical error? I'm curious what you think. I do think that there is a very real tension (even inconsistency) between the two approaches and I have not seen this fleshed out in any books or in any investing forums.
I'd love to hear your thoughts.
That’s a great question.
If you look at the two approaches you can see they're both talking about earning power. What matters is future earnings. But they both tackle it obliquely.
Because you can’t directly measure what hasn’t happened yet. So Graham and Greenblatt each give us indirect measurements of earning power.
They measure something that exists today. Something they think is related to future earnings.
This is the unspoken assumption underpinning each strategy.
Graham's approach is to look at all businesses as if they are generic. There are no special businesses. He's looking at every business as if it is average.
Greenblatt's approach is to look for special businesses. Situations where the normal rules do not apply.
We can think about this like looking at a population. There are special groups hidden among that general population.
Both investors are hunting for future earning power. They're just hunting in different places.
Let’s stick with this hunting analogy.
Say there are both wolves and cougars in the same area. The wolves hunt by smell. The cougar hunts by sight.
The wolves smell something from a mile away and move in slow. They get seen further out. But they outlast their prey.
The cougar zig-zags cross the area. Stays in cover. Stalks its prey up close. Hopes never to get seen till it’s too late. No outlasting. Just ambushing.
Okay. Clearly different strategies, right?
But they could be hunting the exact same prey. They could both be picking off that same sickly deer. Competing for the same kill. Getting the same outcome.
Just using two different strategies to get the same result.
That looks kind of paradoxical.
But it also looks wrong.
Are Graham and Greenblatt hunting the same prey?
They aren’t generalists.
Both approaches — as actually applied — focus on throwing out 90% of the investment world and keeping 10%. Both are disciplined. Way more disciplined than most actual investors.
And both approaches have merit. I would not apply Greenblatt's approach to picking specific stocks. I don't think the magic formula is consistent enough. Even Graham's approach is greatly improved when you use the F-Score.
As Warren Buffett has said: A stock is worth its future cash flows.
How much cash will you be able to take out of the business?
That’s the question both Graham and Greenblatt are asking. But they’re coming at the same question from different angles.
Graham's approach is not exactly low price-to-book. If you look at investment history before Ben Graham's time, book value was paramount. We lose perspective on this today. Graham seems to be over-stressing book value. In fact he was revolting against the use of book value alone.
He also revolted against the idea of focusing solely on dividends. Before Graham's time book value and dividends were important in setting the price of a stock. Graham stressed the importance of earning power. And the importance of net current assets. Graham put new twists on old ideas.
Dividends were important. But earning power was more important. Book value was important. But net current assets were more important.
Greenblatt's approach makes sense up to a point. The problem with the magic formula is that it focuses on a bottom-line number. Earnings are the result of countless interactions higher up in the business.
We’re talking about sales, gross margins and operating expenses. Greenblatt doesn't talk about the variability in each of these items. When I’ve talked about magic formula type stocks — I’ve stressed the importance of the price-to-sales ratio, the free cash flow margin and the variation of that margin. I think that is a better approach for picking specific “magic” stocks.
And I don't think Joel Greenblatt would disagree. The magic formula was created for picking a basket of stocks. Not for picking individual stocks. It actually does not do well in anything but a group approach.
This is similar to picking low price-to-book stocks without applying Piotroski's F-Score. There is a high rate of catastrophic loss among both low price-to-book stocks and magic formula stocks.
If you use a magic formula sorting mechanism and apply it to the entire universe of net-nets you end up with superior performance. A magic formula sorted group of net-nets does better in back tests than a randomly selected group of net-nets. However, you also have a greater chance of buying net-nets that result in huge losses.
Unlike Piotroski's F-Score, Greenblatt's magic formula does not move the entire range of outcomes into more positive territory. The F-Score results in more big winners and fewer big losers. The F-Score cuts the risk of really big losers.
This is not true of the magic formula.
With the magic formula you have home runs and you have strikeouts. You don’t get many walks or singles. You don’t have a very high on-base percentage.
That’s different from an F-Score sorted group of low price-to-book stocks.
I think the magic formula is a poor screen. It works in group operations. It's not a great list to work from. Unless you are very selective. Unless you are looking at the stocks qualitatively rather than quantitatively.
But I wouldn’t just pluck a couple names from the magic formula hat. That’s a risky approach.
One question here is why the two approaches work at all.
Let's start with price-to-book. The price-to-book ratio works over large groups of stocks. There is no doubt about this. Many studies show it.
Use of the Piotroski F-Score improves the results of a pure price-to-book screen tremendously. Without applying the Piotroski F-Score a group of low price-to-book stocks will have more underperformers than overperformers. However, the group as a whole will still outperform the market.
Why does this happen?
One argument is that low price-to-book stocks tend to be in a weak financial position. They’re risky. That’s why investors avoid them.
But why do the survivors perform so well that the group as a whole outperforms?
We won't delve too deep into investor tastes here. Maybe investors are overly risk averse. Maybe a few rotten apples spoil a bargain barrel in their eyes. They don't realize the entire barrel is valuable because they remember only the bad experiences they had with those couple rotten low price-to-book apples. That's possible.
But we want to talk about why low price-to-book stocks perform so well as a group. Basically we’re talking about reversion to the mean. We’re saying that a certain amount of tangible capital is going to tend to earn a certain return throughout the economy.
This is the opposite approach from Joel Greenblatt's magic formula. We don't look at earnings. We just look at the capital in the business.
Let's use a baseball analogy. You probably read Moneyball.
You may remember a chapter in which they talked about valuing pitchers. The big breakthrough was valuing pitchers based solely on their strikeouts, walks and home runs. It made no difference who the pitcher was if the ball was put in play.
This was not the traditional approach to analyzing pitchers decades before.
The idea here is that when a ball is put in play it's a roll of the dice. We should not get worried when we see statistical flukes from one year to the next.
Pitchers will not be permanently lucky. Or permanently unlucky.
Companies will not be permanently lucky. Or permanently unlucky.
That is the Ben Graham approach. That is the low price-to-book approach.
But notice that there are some things pitchers can control. They can control walks. They can control strikeouts. And they can control home runs.
This is the Joel Greenblatt approach. Some things are random. But not everything.
The Ben Graham approach works because it takes advantage of prices that reflect nothing but luck. Companies that earn a lot today may not earn a lot in the future. Companies that earn very little today may soon earn a lot again.
If return on capital tends to be driven down to the same level getting as much capital as possible for each dollar you pay is the best approach to picking stocks. Low price-to-book wins.
But is the business world really just one big cattle drive?
Are there no abnormal companies?
No odd ducks?
Just faceless capital getting bumped about but ultimately destined to move at the same economy wide speed?
Ben Graham never said that all companies are equal. He never said that there weren't great companies. He never said a company couldn’t keep earning a high return on capital.
Ben Graham said that wasn’t his game. He was never comfortable paying much more than book value for any stock.
Graham knew intangibles were valuable. He just didn’t feel comfortable valuing them himself.
Phil Fisher took the opposite approach. He focused on finding unique individuals in a seemingly faceless herd. Finding those few companies that can earn consistently high returns on capital now and forever.
The Graham and Fisher approaches look totally opposite. They aren't. They are just exclusive. They don't overlap. They’ve got nothing in common.
And a lack of common ground isn’t necessarily a contradiction.
You could just be speaking on points the other guy is silent about.
Ben Graham is looking at ordinary companies trading at extraordinary prices. Phil Fisher is looking at extraordinary companies trading at ordinary prices.
The Phil Fisher approach can't work in an environment like the late 1990s. You can't pay 100 times earnings and still make money in a great company. But as long as investors keep P/E ratios in a reasonable range you can make money solely by picking the most extraordinary companies.
This bunching up of P/E ratios is why both the Graham and Fisher approaches work. Today's earnings are not a good metric for valuing all companies.
A low P/E approach may be slightly better than randomly picking a stock. Some studies show this.
Buying or selling the market based on the current P/E ratio does not seem to improve performance. Using other metrics does. Long-term average P/E ratios like the Shiller P/E ratio do work for “timing” the market. And the price-to-book ratio works too.
The market tends to extrapolate the current situation into the distant future. It is myopically focused on today’s earnings.
This is different from Ben Graham's day. When Ben Graham started on Wall Street the market was myopically focused on dividends. In the 1920s the focus shifted to earnings growth.
So we should not set down rules that apply for all times. Rather we should say that markets sometimes over apply one rule to all stocks.
By seeing the market’s blind spots we can make money. But markets adjust. Environments change. There are investment revolutions. The conventional wisdom changes. So we cannot say that one approach will work in all places and times.
For many decades now the market has been preoccupied with earnings. All you hear on TV is what the market’s price-to-earnings ratio is.
Do you hear about price to sales? Do you hear about price to tangible book?
Why is this a mistake?
Because profit margins are not permanent. Returns on equity fluctuate.
Go back into the history books and look at what the return on equity was in the 1940s and 1950s. Now look at what the return on equity is today. Keep in mind that long-term interest rates are not all that different. Long-term corporate bonds had low yields in both cases.
So why should we assume today's very high returns on equity will be sustained?
That’s the lesson Ben Graham teaches. It is a mistake to pay too much for each dollar of capital a stock is backed by. It isn’t today’s earnings that matter. It is future earning power.
For most stocks, earning power is closely tied to book value.
But you can take this too far.
Many people do.
It makes no sense to talk about Coca-Cola’s price-to-book ratio. It makes no sense to talk about FICO’s price-to-book ratio. It makes no sense to talk about Dun & Bradstreet's price-to-book ratio. And it makes no sense to talk about Microsoft's price-to-book ratio.
Because you can't replace those businesses.
That capital is special. It is not fungible.
Those companies will earn high or low returns based on intangibles. They can become impaired. But their returns will not drop to reflect the low amounts of tangible capital. Their returns will drop if and only if their unique intangibles erode.
The price to book ratio works to the extent that capital is fungible. But capital takes different forms. It does not easily change from one form to the next. Some capital is more flexible than other capital.
Ben Graham saw this. He knew that current assets provided a better margin of safety than property, plant and equipment. Fixed assets are hard to repurpose. Cash is not.
Ben Graham's approach is not a strict price-to-book approach. His approach reflects the special nature of some capital. He knows that not all capital is equally valuable.
Greenblatt goes to the opposite extreme. He mechanically applies a Fisher like approach. While it appears this is the opposite of Ben Graham's approach it really isn't. It's a different hunting ground entirely.
Ben Graham looks for low prices. Joel Greenblatt looks for high quality companies.
Greenblatt's unspoken assumption is that these magic formula companies are not just lucky. Something is causing them to earn such extraordinary returns on capital.
Some of them are insulated from competition. And now we are edging closer to the Warren Buffett approach. Warren Buffett would not approve of the magic formula’s mechanical approach. But he would approve of starting with a list of magic formula stocks and then finding those few companies with wide moats.
The problem is testing this. How do you define a wide moat?
Academics can discuss the Ben Graham approach. And the Joel Greenblatt approach. They can't really talk about the Warren Buffett approach.
But we can.
Let's try a thought experiment.
Let's talk about baseball again. My apologies to any Europeans out there.
A magic genie comes to you. The genie says I will give you $10 million if you can pick the best batter in baseball. I will give you $5 million if you can pick one of the two best batters in baseball. And I will give you $1 million if you can pick one of the three best batters in baseball.
I'll make it easy. I will let you pick one batter for every five batters in the league. You get to pick 20% of the league.
If one of the 3 best batters is in that 20% you win.
Pretty easy, right?
Here's the catch. I won't let you see their names. And I won't let you see all their stats. Just one stat. I will let you see their strikeouts. That's it.
Now those of you with a limited knowledge of baseball probably think strikeouts are bad. That's true. They are.
A baseball game has nine innings. And no time limit. Each team gets three outs per inning. So the goal is to score as many runs as possible using your 27 outs.
A strikeout uses up an out. And does nothing productive. So a strikeout is bad. No argument there.
Does that mean you should pick the quintile with the lowest strikeouts?
Turns out that’s an awful strategy.
Strikeouts are bad. Batters who strikeout are not bad.
Some of the more statistically inclined among you might have a different idea. Maybe strikeouts don't matter. So I shouldn’t just pick the 20% in the league with the lowest strikeouts. Instead I should take a random sample from the group. But this is really a forfeit. A cop out.
And it’s not the ideal strategy.
Let’s think this through.
The genie gave you a list. He gave you an advantage. You’ve got info. And yet you're giving up.
Because you think you don’t know how to use the limited information you’ve got.
But even if you don’t fully understand strikeouts the list still tells you something. In fact it tells you three things.
It tells you which players strikeout a lot. It tells you which players strikeout a little. And it tells you which player strikeout a normal amount.
We are looking for extraordinary players.
Does that mean we are looking for players that strikeout a lot? Does it mean we are looking for players that strikeout a little? Or does it mean we are looking for players that strikeout a normal amount?
Which question is the easiest to answer?
I say it's the third question.
Are we looking for players that strikeout a normal amount?
The answer is no.
We are looking for extraordinary players. If extraordinary means anything it means not ordinary.
The one group we do not want to sample from is the group bunched up at the average. We do not want the median player. We do not want to skim anything off the top of the bell curve.
We need to go out the curve.
Even before we know whether we should look left or right we know we should not look at the place where everybody’s bunched up.
We are not looking for a normal player. We are looking for an abnormal player.
We can take the decile of players who strikeout the most. And we can take the decile of players who strikeout the least.
We are still agnostic about whether strikeouts are good or bad. But we are not agnostic about the fact that striking out tells you something.
What does it tell us?
Batters have a lot of control over strikeouts. If a batter strikes out a lot he is doing something differently. He has an abnormal strategy. A different routine.
That's interesting. It draws our attention.
We want to find a special group among the general population. We want to focus in on a group that holds more promise of having the most special player of all.
Note that this does not mean the group we mark off from the general population will necessarily be better or worse on average than the league. The average member of the group is not our concern.
Our concern is making sure that the most unusual player in a good way is in the group that we pull out from the general population.
I know this sounds strange.
But it could be a mistake to carve out a group from the general population with an eye on the average of that group. We want to find special players. So what matters is not necessarily the goodness or badness of the group. What matters first is the specialness of the group.
Are we getting a special group?
Is it unlike the general population?
That's what we are most concerned with.
And that's the first thing that any investment strategy has to look at. To be better than average it must first not be average. The strategy must be different.
Now let's look at the actual players who did strikeout a lot. I have in front of me a list of the 44 most strikeouts in a year by any batter.
Note that this list does not have anywhere near 44 players on it. This shows that batters have a lot of control over strikeouts. The same players show up on the list year after year.
In fact the player who holds the record for the most strikeouts in a single season also has the second most strikeouts in a single season and the third most strikeouts in a single season.
The average of those three record setting strikeout seasons actually results in an on base percentage that's quite normal for the era.
Which is weird.
And throws doubt on the whole strikeouts are bad idea. Logically we know strikeouts must be bad. But empirically the situation is looking doubtful.
We also have some players on this list like Adam Dunn who have good seasons with a lot of strikeouts and bad seasons with a lot of strikeouts.
The overall list has a high number of very good players on it.
You have years like Ryan Howard's 2007 season. He struck out 199 times. Only a handful of players have ever struck out more in a single season. In that year Ryan Howard had an on-base percentage of .392 and a slugging percentage of .584. He came in fifth in the MVP voting.
Clearly strikeouts are not all bad. But strikeouts themselves are bad.
Is this a paradox?
There's no denying that a strikeout is worse than no strikeout. And there’s no denying that book value is better than no book value.
Perhaps we’re being too simple.
Batters are complex. Businesses are complex.
We may be measuring a part. And ignoring the whole.
We are looking at one tiny aspect of a complex system. Something where there is strategy. Where there is a routine.
Using a strategy that involves a high number of strikeouts often works well. Batters who strikeout a lot are often very good batters. So risking strikeouts may be part of a good strategy.
Maybe you get to pick pitches. Maybe you walk more. Maybe you hit more home runs. Ironically striking out seems to be part of a strategy that involves getting safely on base.
It’s not the only strategy. There are others ways to get on base. No one is saying strikeouts are necessary. Just that they do happen as part of some good approaches to hitting.
So strikeouts don’t seem to measure what we think they measure. The stat doesn’t matter in the perfectly linear way we like to think things matter.
Low good. High bad.
It doesn’t work that way.
Let's apply this to the corporate world. Normally having a lot of tangible equity per share is a good way to ensure a lot of earnings per share. A company with tangible book value of $100 a share is more likely to earn $5 a share than a company with $20 a share of tangible book value.
But aren’t there companies that can earn a lot on almost no tangible equity?
Warren Buffett owned both Moody’s (MCO) and Gillette when they had negative tangible book value. I bought IMS Health when it had negative tangible equity.
Those investments worked out.
And if some companies can earn money on literally less than no tangible equity — the price-to-book ratio must be meaningless in some situations.
Equity is something. But it isn’t everything.
Generally price-to-book works. But there are special exceptions.
This is the reason for the magic formula’s successes and its failures.
Where the magic formula fails it fails because it accidentally lumps lucky companies in with special companies.
Some normals have slipped in among the abnormals.
That's always going to be a problem for the magic formula.
It's also going to be a problem for the price-to-book ratio. But the other way around. Price-to-book misses the special stocks. It ignores the abnormals.
If you’re using a long only approach this isn’t the worst thing in the world.
If you’re shorting stocks purely based on price-to-book value — it gets dodgy.
The Ben Graham approach works well over large numbers of stocks. But Ben Graham himself gave up shorting specific stocks.
Why is going long low price-to-book stocks and going short high price-to-book stocks not something Ben Graham kept doing?
He decided it was more trouble than it was worth. A review of his performance showed shorting stocks did not improve his results.
Almost certainly because he accidentally shorted special companies as if they were normal companies.
Look at the case for shorting a stock like Netflix (NFLX) or Amazon (AMZN). It's really very simple. The bulls are arguing this business is special. It doesn't follow the same rules most businesses follow. The bears say competition is coming. This is just a lucky run. The stock price has to come back down to earth.
It's just a disagreement about whether we are dealing with a company that is extraordinary or ordinary.
Should you short Coca-Cola (KO) when it has a high price-to-book ratio?
Should you short Moody’s (MCO) when it has a high price-to-book ratio?
Moody's does not require one iota of tangible equity to earn a lot per share.
There are many reasons to short Moody’s. Price-to-book isn’t one of them.
The reason why the price-to-book ratio of Moody's or other similar businesses doesn't matter is that you are no longer comparing the price to anything meaningful. Book value doesn’t matter at Moody’s. So Moody’s price-to-book ratio can’t matter.
But is the price-to-sales ratio still meaningful?
Moody's can only squeeze so much free cash flow from every dollar of sales.
At 10 times sales it would be overpriced.
So an astronomical price to sales ratio would indicate that Moody's is overvalued. The price to sales ratio still has meaning. Even when the price-to-book ratio no longer has meaning.
No one is saying that price itself doesn't matter.
Price always matters.
Sometimes book value doesn’t.
It’s always best to keep the idea of earning power in mind.
Earning power means a certain return on equity or a certain return on sales. You need to tether earnings to something grounded in the business. Otherwise earnings is just a meaningless abstraction.
If a company is supplying an essential product and it has few or no competitors then you may be able to assume very high prices. And very fat margins.
Those businesses are rare. That's why the price-to-book ratio often works well.
The answer to this paradox is that special groups hide among the general population.
It matters what group we’re looking at.
Are we dealing with normal or abnormal businesses?
It's not really a paradox.
It's actually a case of not breaking up the general population into smaller groups.
Academics haven’t done this.
Unfortunately these special groups of companies don’t lend themselves to study using the kinds of metrics academics focus on. They aren't able to isolate the special groups in the general population of companies.
But there may be ways of doing this.
I've talked about the coefficient of variation for a company’s free cash flow margin.
Because I think it is one clue in identifying a special group.
High, stable margins suggest a business is special. High, stable returns on tangible assets suggest a business is special.
These are clues that the price-to-book ratio may be misleading for those stocks. These are clues that you're dealing with a special business.
That's as far as I've gotten quantitatively.
Beyond that I haven't found any clues that can be put simply in numbers.
You really do have to take the Phil Fisher approach when studying special businesses.
Follow Geoff at Gannon On Investing