Someone emailed me a question about Activision (ATVI), a stock I put 100% of my portfolio into a little over 16 years ago (the stock went on to return 22% a year – but, of course, I didn’t hold on to it these last 16 years):
“Would it be fair to say that your returns would have been much better had you just put all your money into Activision at the time you initially bought it… and just sat on your butt until now? Let's assume that this is a fair assessment for now.
So if we brought ourselves back to the year you bought it, early 2000s was it? If we looked at it with the models you currently possess but likely did not possess back then, could you have made a better allocation based on those models alone?”
The only “model” I can think of that would have improved my performance is not letting myself make any conscious sell decisions. In other words, just selling pieces of all the stocks I own in proportionately equal amounts to fund new purchases, never selling just to hold cash, etc.
I wrote an article discussing some of this. Overall, my sell decisions haven’t added much (if any) value to my investing record. My investment results are primarily a result of taking larger than normal positions in some stocks and then secondarily in picking the right stocks more often than I pick the wrong stocks.
With hindsight, I would have done as good or better while doing far less work if I’d just stuck with a stock like Activision that I once (16 years ago) had the conviction to put 100% of my net worth into it.
However, I think there is both: 1) A valuable truth and 2) A dangerous falsehood in this kind of thinking. Basically, what you’ve uncovered here is a good idea. But, a good idea can be taken to a bad extreme. And, I think the combination of 1) abusing hindsight and 2) going off the stock performance rather than the business performance can skew just how good and certain an idea Activision really was in September of 2001 (when I allocated 100% of my portfolio to it).
I couldn't have foreseen that Activision would return something like 20% to 25% a year for the next 15-20 years. At the time, I thought I was able to foresee Activision could return 10% to 15% a year for the next 10-15 years though. Now, it’s true I thought this thought with enough “certainty” that I was willing to put 100% of my portfolio into the stock. But, I didn’t go “all in” on Activision believing I could make 20% to 25% a year. I did it believing I could make 10% to 15% a year (with greater confidence than I had in any other stocks).
Since 2001, Activision’s capital allocation has turned out to very good, or very lucky - or some combination of the two. Should I have known that would happen? To some extent, yeah. A key reason – really, the key reason – for me buying Activision instead of something like EA is that I liked Activision's management and capital allocation a lot better. In fact, I disliked EA’s management so much I’d never consider buying the stock as long as that management team was there. And I really liked Activision’s management a lot. It’s not like I was neutral on the top people there. I thought they were saying the right things about capital allocation at a time (around the turn of the millennium) when no one was saying the right things about capital allocation. I knew that in a business like video game publishing, book publishing, movies, etc. you bet the “jockey” if by “jockey” we mean best capital allocator. That’s because these businesses produce tons of free cash flow, so your return is largely going to be the return on re-invested cash. There’s no requirement to put the cash back into the franchise that earned in. In fact, you often can’t do that. A hit media franchise can never come close to re-soaking up all the free cash flow it gushes.
Okay. So, I should have held Activision longer. Is this part of a bigger pattern for me?
In general, I haven't been much better off selling a stock I bought to buy another. Certainly, of the stocks I sold within about 10-15 months - where things were going well and the price was rising - I would have been better holding for 10-15 years.
You can see this even more recently. For example, I bought FICO about 7 years ago. The stock has returned over 25% a year since then. I didn't hold it from then till now.
What you have to be careful about here though is multiple expansion. When you buy something like Activision, J&J Snack Foods, FICO, or Village Supermarket (these are all stocks I bought a lot of when they were really cheap) at a low multiple of sales, book value, earnings, etc. you can consider the substantial annual return bonus you get in the stock from multiple expansion to be one-time but fully justified (and therefore not going to be reversed in future years) up to a point.
I’m going to spend the rest of this post explaining what “up to a point” means. It’s one of the most important concepts to long-term, buy and hold investors. This idea that you are – if you buy only cheap stocks – entitled to getting one and only one multiple expansion “bump” to your returns is something buy and hold investors need drilled into their heads during the last stages of a bull market. Since we may now be in the last stages of a bull market, let’s talk about how a justified initial multiple expansion in a stock can quickly morph into a totally unjustified subsequent multiple expansion.
Let me give you some examples.
Activision: I bought this stock at something like an EV/Revenue of 1. It now trades at an EV/Revenue of more than 6. This multiple expansion counts for over 10% of the annual return in the stock. How much of it is justified? Activision shouldn't trade at an EV/Revenue of 1, but I'm not sure it should trade at an EV/Revenue of 6 either. Today, the stock would have to fall more than 50% before I'd say it was clearly "cheap".
FICO: I bought this stock at something like an EV/Revenue of 2. It now trades at an EV/Revenue of more like 5. This multiple expansion counts for over 22% of the annual return in the stock. FICO might have to fall close to 70% before I'd say it was clearly "cheap".
You want to be careful about this, especially as we are now in one of the longest lasting bull markets ever. It's often better to look at your returns in terms of the business results than the stock results. So, judge your returns by the increase in per share sales, free cash flow, etc.
Having said that, you must also take justified multiple expansion into account to some extent if you're a value investor. I bought Village Supermarket (VLGEA) at a P/E of let's say less than 7 and an EV/Revenue of around 0.1. It was reasonable I think to believe that because the business was a perfectly decent one it would eventually deserve a P/E of about 15 and an EV/Sales of about 0.2 or 0.25. Beyond that, you are starting to get speculative. I often think in terms of what I think I can get as a return over the next 5 years if the company's stock takes that long to get valued at what I think is the "right" multiple. For me (a value investor) this means I am usually buying below the market's "normal" P/E and expecting the stock to at least rise to that level. So, I’m buying Village at a P/E of 7 and expecting it to one day trade at a P/E of 15. For some exceptional businesses – like Omnicom, FICO, and BWX Technologies – I may be buying at a P/E of 15 and expecting the stock to one day trade at a P/E of 25. What I’m not doing though is buying at a P/E of 20 and expecting the stock to one day trade at a P/E of 40 – even though, I know, there may come a time where Mr. Market gets overexcited with this kind of wonderful business and really does give it a P/E of 40.
So, does that mean I should sell when it reaches that level? If I expect a multiple expansion from a P/E of 7 to a P/E of 15 for an average business or from a P/E of 15 to 25 for an above average business, should I sell the second a stock I own hits my P/E “target”?
But, even if you look at someone like Buffett's returns - you can see two features. One, he often did fine if he never sold. This is true even of stocks he did sell. For example, Buffett bought General Dynamics (GD) shares in the early 1990s. He sold the stock. But, he would've done very well if he'd hung on to General Dynamics for the last 25 years. Why? Probably because Buffett analyzed the business and saw it was a good one and he saw that capital allocation was going to be good and then the CEO then in place followed through with that kind of capital allocation and the CEOs that followed him copied those practices. So, the two things Buffett saw that he liked: good economics and good capital allocation proved to be durable.
As an example of what’s durable here – the industry structure of defense contractors and the market power they have with respect to their customer (the U.S. government) and their suppliers (often companies who do not deal directly with the U.S. government the same way they do on big projects) is going to tend to stay the same. If market power and capital allocation don’t change – the right business to own in the 1990s will often stay the right business to own in the 2000s, the 2010s, etc.
This is why it's often a bad idea to ever sell a stock you've bought. In a sense, you are making a very tough bet to get right. You're saying that you correctly judged the quality of the company to be high, its future to be bright, etc. but now you are correctly judging that the company's quality, future prospects, moat etc. are not high enough to overcome the current elevated price. That's a tough bet to ever win. The more certain you were of this business in the first place, the more you should doubt ever selling it. This is if you did the sort of in-depth look at moat, etc. that someone like Buffett does or that I often do. If I think WTW has a moat (like its relative size versus competitors), it's hard to later be right in saying the moat has been breached. If I think BWXT’s industry structure (it’s a monopolist in some areas and an oligopolist in others) and product economics (it rarely has to tie up much capital ahead of time in anything it does) are so favorable it’s the right stock to buy at a P/E of 15 – then, it’s much harder than you might think for me to reverse myself correctly at a P/E of 30 from a handicapping perspective. This requires an ability in precise quantification that I probably lack. How much is the right product economics and the right industry structure worth? It’s hard to say. What I know is I researched these questions in-depth and liked the answers when I first bought the stock. To do what is probably a more superficial (and biased towards “recency”) analysis later that overturns my initial decision – that’s hard to do. Common law legal systems operate on a principle of precedent. If you believe you made a correct precedent setting decision years ago, leave that decision undisturbed now. A lot of investors are making decisions based on beliefs they might have 50% to 75% confidence in. Don’t do that. You want every decision you make – including overturning a prior decision you made – to be decisions you make with 90% to 100% confidence. The likelihood that you’ll ever have 90%+ confidence in a decision that overturns a previous decision you made is extremely low. Who can reverse themselves with 90% confidence?
Okay. So, whenever possible, leave your stocks undisturbed and your prior decisions in place.
But, I did say there were two features you see with the buy and hold practices of someone like Buffett.
The other feature is that while it's true that if you measure from the time he bought a stock till the time he is still holding it decades later, the return is still often good despite him not selling out at an earlier date - the annual returns is often not as good as it once was. This isn't always true. For example, General Dynamics did worse over the first 10 years from the time Buffett bought it versus the subsequent 15 years. However, if we break down the returns in some of Buffett's favorite stocks into two time periods: the first 10 years and then all the other years - we can see the first 10 years sometimes had really amazing results.
Based on information in Berkshire's annual letters, I estimate Buffett's Washington Post stock returned about 32% a year for the first 10 years he owned it. The stock still did fine as an investment for Buffett once you include the next 20-25 (taking it through the 2000s). But, the returns were very strong in those first 10 years.
Coke shows a similar pattern. The return in that stock was something like 25% a year for the first 10 years. That stock hasn't done well over the last 18 years. A lot of people will blame changes in consumer tastes for that. But, really it was Coke's P/E. The company experienced multiple expansion for the first 10 years Buffett owned it that was very extreme.
This is why people should be careful about the FANG stocks. Not because they aren't great businesses, but because their stock prices have been growing faster than the underlying business value.
Facebook: Over the last 5 years, there's been a 5% boost to the annual return in FB stock due to multiple expansion (EV/S).
Apple: Over the last 5 years, none of AAPL's annual return has been due to multiple expansion.
Amazon: Over the last 5 years, there's been a 12% annual return boost in AMZN stock due to multiple expansion.
Netflix: Over the last 5 years, there's been a 43% annual return boost in NFLX stock due to multiple expansion.
Google: Over the last 5 years, there's been a 7% annual return boost in GOOGL stock due to multiple expansion.
I based all those multiples on sales, which is usually the safest way to do it. Investors often use the P/E multiple. That can be risky though. Because you are then assuming that both higher sales and higher margins (today vs. the past) are normal. Unless you have very strong evidence about the long-term structure of this business as it scales, I would strongly suggest using growth in sales per share or maybe growth in gross profit per share rather than growth in EPS as your guide to intrinsic value growth.
I actually looked pretty hard at Netflix about 5 years ago and couldn't bring myself to buy it because I'm too much of a value investor who focuses on certain multiples you get accustomed to paying in terms of earnings and things like that. But, Quan and I both knew Netflix was cheap and it was going to have a wider moat in 5 years than it did in 2012.
Should I blame myself for not having the flexibility to break free from some of that value investing dogma that fills my mind and buying Netflix?
But, should I blame myself for missing out on an 80% annual return in the stock?
No. Netflix has only grown its revenue per share by something like 20% a year.
Two things have happened to expand Netflix's share price result far beyond this. One, it grew debt. Two, investors valued each dollar of sales higher (even when it came attached with more debt).
A re-rating of Netflix from having an EV/Sales of 1 to having an EV/Sales of 2 or even 2.5 might make sense based on the company's own past history. Knowledge of the historical economics of similar media companies (cable networks, TV stations, etc.) might even get you to a belief that if Netflix would become both dominant and mature it might even deserve a EV/Sales of 4 at that time. However, no analysis I'd be able to come up with would ever tell me that Netflix deserved an EV/Sales anywhere near 8 unless it was still growing phenomenally fast.
The danger here is always that because of the combination of a strong business performance and then a strong stock performance on top of that (due to multiple expansion) we may become more sure of Facebook, Amazon, Netflix, and Google or of Activision or FICO or whatever long-term winner we own in our portfolio. And the truth is that while some of those businesses are certainly on more stable ground today than when I first analyzed them - other aren't. FICO isn't a wider moat, better company today than when I looked at it. FICO should have been priced at the exact same multiple of sales in 2010 and 2017. Instead, Mr. Market is willing to pay 2.5 times more per dollar of FICO’s sales in 2017 than he was in 2010. FICO was too cheap in 2010. And it’s too expensive now. I could’ve gotten a tremendous return – something like 25% a year – in FICO by holding all the way from when I bought it (knowing it was way too cheap) in 2010 and holding it till today (when I know it’s way too expensive).
But: Is that smart? Is it safe?
Nothing happened with FICO that I didn’t pretty much expect to happen over the next 7 years – except for one thing: I never expected the stock to end up with a P/E of 40.
Let’s look at Netflix over the last 5 years. I'm not sure I'd say Netflix's competitive position today is different from where I expected it to be at this point in time. However, the stock is probably 4 times higher than where I would've told you it should be. In other words, I could have correctly – in very rough terms – guessed where Netflix might be in terms of its number of subscribers, how much it was charging, and how much competition it was facing for the acquisition of content (this was always my biggest concern). Now, I was nowhere near 100% certain of my guess as to where Netflix would be in 5 years. Otherwise, I would have bought the stock. But whether I was 51% certain or was I 75% certain of where the business would be – I still never would have guessed where the stock would be today. I’d have guessed that if Netflix hit all the expectations I had for it (as a business) the stock might return 20% to 30% a year over the next 5 years – not 80% a year.
If you look at Buffett's investment in Coke, the P/E on that stock expanded by about 13% a year in the first 10 years he owned the stock. And I'm sure Buffett would say the quality of those earnings deteriorated as well. Now, I don't mean to say that Buffett was wrong buying Coke at a P/E of 15-17 or whatever he bought it at and believing it deserved a P/E of 25 sometime down the road. We can see from the stock's subsequent history that outside of moments of real financial stress in the market - Coke hasn't really had a P/E of 15-17 since Buffett bought it almost 30 years ago. So, he was justified in the belief (if he had any such belief) that Coke deserved an expansion of its P/E ratio. But, Buffett wouldn't be justified in believing Coke deserved a re-rating in the P/E from 15-17 to say 45-51 (3 times expansion). And yet, at times, Coke actually traded at such a P/E and the stock's long-term performance would have looked amazingly good during those periods.
That brings us to the classic question: Should Warren Buffett have sold Coke in the late 1990s?
He has a different calculus than the rest of us. Berkshire brings in a lot more new cash to invest all the time. So, I think by not buying more of stocks he already owns he is watering them down in much the same way an individual investor would be when he sells all the stocks he owns in equal proportions.
This is the approach I've chosen to take.
You can see that with my latest purchase. At the start of October, I put 50% of my portfolio into one stock. However, I only had about 30% of my portfolio in cash at the time. So, I had to sell something. Instead of selling all of BWXT (the most expensive stock I owned) I sold about one-third of BWXT and one-third of Frost which means I express no preference when it comes to selling. Now, actually I prefer Frost over BWXT at today’s prices. But, I forced myself to ignore that. I’m trying to only express my preferences in buy decisions – not sell decisions, from now on. Basically, I’m trying to “buy right” and then just forget about what I own.
If I continue to apply this rule, it means I will slowly sell down stocks I own over several years making them a smaller and smaller part of my portfolio as they age.
For example, I took a 50% just now. If I buy a new 20% in 2018 and another new 20% in 2019 and then another new 20% position in 2020, and so on…
That would hypothetically (if the 50% position had the same performance as my other stocks) result in the stock I just bought being 50% of my portfolio in 2017, 40% of my portfolio in 2018, 32% of my portfolio in 2019, 26% of my portfolio in 2020, 20% of my portfolio in 2021, and so on.
That’s not a true “buy and hold” approach. But, it both relieves me of having to make sell decisions and yet also gradually clears out my old ideas (which have presumably risen closer to their intrinsic value) and replaces them with my new (and hopefully cheaper) ideas.
This is the approach I think makes the most sense. I think buy and hold makes a tremendous amount of sense and I recommend it to people who have a constant influx of cash into their savings and don't have a ton of time to ferret out new stock ideas.
In fact, for the average stock picker my advice is:
· Take everything you saved this year
· Buy just one stock with all those savings
· Never sell that stock
· Repeat every year
That might lead you into situations like Activision and lead you to hold them forever. It would certainly cause you to be focused on what I think matters most: your highest conviction buys.
Superficially, it would also seem that this approach should lead to wide diversification. However, in practice, this is unlikely to happen. Your winners will become much bigger than your losers if you truly never sell the winners.
I think it's important - especially as I write this in October of 2017 - to consider how with hindsight the results of buying and holding the right stocks through a bull market look better than we should perhaps expect we can do in the future. There are stocks that may have looked like Activision and didn't work out. And then there are periods (like 1965-1982) where multiples simply do not expand on stocks.
However, have I often sold too soon?
Yes. There are definitely chances I missed to buy and or simply hold a business I knew well and liked. But, I'd also say that much of the subsequent return in these stocks is something I couldn't have counted on (an expanding multiple beyond the historical norm).
I think it's reasonable to buy an undiscovered or misunderstood stock at a low multiple and expect a one-time re-rating of the correct price multiple as that stock is discovered and better understood by the public. However, this is a one time and one time only bump in the stock that may take 3, 5, or even 10 years. A multiple expansion from an EV/EBITDA of 6 to 12 may be justified in cases where you know just how great a business is and the market doesn't yet. An expansion from an EV/EBITDA of 12 to 24 won't be justified ever. But, it will still happen to some stocks you own and while you own it this second expansion may not feel that different from the first (fully justified one).
The thing about bull markets both in the overall market and in specific stocks you own too is that a good idea is first latched on to by a few very smart people and then over time some less and less intelligent people doing less and less in-depth work of their own on that stock take this good idea and they take it way too far.
When I tell the story of BWX Technologies (BWXT) it makes a great deal of sense at half of today's price (about where I bought it). But, that same story makes a lot less sense as a reason to buy the stock today. The story is the same: BWX Technologies is still a high return, monopoly business that can pass inflation along to its customer (the U.S. Navy). It's clear that you should buy such a business at a P/E of 15. At a P/E of 30, it's less clear. And yet it seems more certain (because of the stock's multiple expansion) to people looking at the stock today than when I first wrote about it.
That's the fear I have when talking about some of these businesses. What you want is to buy an above average business at a multiple below what will be justified in the future when the stock is still undiscovered or misunderstood.
When I ask for examples of great businesses from people, they bring me examples of businesses that are already recognized by the market as being great.
If you read what I write about the stocks I own – Frost, BWXT, etc. – or stocks I would seriously consider buying (Omnicom, Howden Joinery, etc.) you’ll notice that I’m saying both:
1) I think this is a great business AND
2) I don’t think the market fully appreciates this is a great business
So, with BWXT I’ll say that unlike with other stocks you have greater visibility into the long-range buying plans of BWXT’s customer in real terms. With the stock at a P/E of 30, this may now be recognized. At a P/E of 15, it wasn’t. With Frost, I always say that this is a far above average bank at a normal Fed Funds Rate. With a Fed Funds Rate at 0% to 1%, I don’t think the market recognizes this. With a Fed Funds Rate at 3% to 4%, it will recognize it.
It’s nice to talk about stocks I bought 15-20 years ago that have worked out well. But, the market recognizes the quality of those businesses today. When I bought J&J Snack Foods (JJSF) like 17 years ago, it wasn’t recognized as being anything other than a mediocre small-cap stock (the P/E was 12). Over the following 17 years, the business changed far less than the stock did. The stock became recognized.
It’s not worth spending even a second thinking about businesses the market already recognizes as great. The problem with FANG stocks isn’t that they aren’t great businesses. It’s that everyone knows they are great businesses.
As a stock picker: Your job is to find a great business no one thinks is a great business yet.