Running PSTG through the Antifragile Framework

Greetings Fools,

Using my Anti-fragile framework, I wanted to show how PSTG stacks up.

Barbell Strategy

Mission Statement
“to deliver data storage that transforms business through a dramatic increase in performance and reduction in complexity and costs. .”

Simple: Yes, if I’m an employee, I can simply ask: “will this (1) increase performance while (2) reducing complexity and cost?”
Optionable: Yes. There’s a whole host of ways to increase performance and reduce cost and complexity – and I’d be willing to bet those ways will change over time.
Inspirational: If that’s your thing.

+1.5 points


Brand: Pure Storage has an audited net promoter score of 83.7 – which is very high. At the most basic level, this means that Pure’s customers are very likely to suggest using Pure to someone else. That said, the company was only founded 7 years ago and FlashArray is barely 5 years old. Therefore, I’m only giving it a half-point.

+0.5 point

High Switching-Costs: The evergreen upgrade-ability without downtime is a big deal. Customers who use Pure Storage and purchase the hardware are much more likely to be sticky. That said, this is still a very young company, and even though the business model is specifically designed to service customers for a minimum decade, the recurring revenue that’s the hallmark of a high switching-cost business isn’t really there yet.

+1.0 points

Network Effect: The company is pushing machine-learning, predictive AI analytics. With anything like this, the more data you can feed the AI, the more accurate it can become over time, and the greater the edge is against competitors who don’t have access to the same database that Pure Storage does. It’s still early in this game, so I’m not giving it two points, but clearly, this tool gets more powerful for everyone involved with each additional customer.

+1.0 points

This is a really tough one for me. I’m not exactly an expert on storage technology. That being said, I will give this one point. There are other products they can provide, and the evergreen state means that applications can be continuously updated.

That said, the big bet here is on solid-state devices – which is the disruptive technology of the day that Pure Storage is betting on. The problem: disrupters can often get disrupted as well.

+1.5 points

Skin in the game

Role of founder: John Colgrove is Chief Technology Officer.
+1 point

Insider Holdings As of the last proxy, insiders together owned 66.5% of voting power.
+1 points

Glassdoor: The company has a 4.0 rating (out of 5 stars).
+1 point

Financial Fortitude

Financial statements
Cash: $551 million
Debt: $0
Free Cash Flow: ($4 million)

This is a really, really tough one to peg. There’s no doubt that Pure Storage has a healthy balance sheet and cash flow statement. But I don’t think it’s so powerful that it could undercut the competition in the scenario of a massive market downturn. It’s simply robust: safe, but not adding any additional long-term strength.

+0 points

Concentration Risk
At the end of fiscal 2016, two channel partners represented 22% of revenue. That dipped to one channel partner representing 11% at the end of fiscal 2017. That being said, it’s probably a safe bet that two still make up over 15% of revenue.

The company also relies on just a few providers for its solid-state devices
-1 points

Total Score: 6.5 points

I’ve run companies through this framework for well over a year, but I do it anew every time (I find that doing so…while laborious…forces me to be more aware of the current situation). As such, I don’t have a huge database to draw from. This lands PTSG near the middle of the list, though it would likely be much higher when compared against all companies.

It’s important to note that I’m nowhere near as “agile” an investor as the average member of Saul’s board. I prefer to buy my companies with a decades long time horizon and – for the most part – following through on that intention. I might make ten to twelve trades in an entire year if I don’t need to free up money for real-world expenses.

The real risk here is commoditization. Pure Storage has the first-mover advantage. The real key is whether the company can build in higher switching-costs and a better AI machine because of this early-mover advantage. That’s no guarantee, but it also means that it could be sitting on a winning lotto ticket if things play out optimally.

See all my holdings here:

What is Antifragile framework and how has it performed?

There are few books that have affected the way that I approach the world quite like Nassim Taleb’s Antifragile. In it he makes a few key observations that have shifted the lenses I use on a daily basis. Some of these things others have said many times, but for some reason, the way it was said stuck with me.

The future is completely unpredictable, and if we use a narrative to try and predict how things will unfold in the world beyond our immediately control, we will likely be wrong.
We cannot reliably measure the likelihood of anything happening.
In the absence of such things, we should measure the fragility of things, which is entirely possible, and eschews relying on any predictions.

In essence, everything in the world can be broken down into three categories

The Fragile: These things will break as a result of time, chaos, stress, pressure, and unexpected occurrences. Think of a china plate set, on the edge of a table, near a train track
The Robust: These things will stay exactly the same, regardless of the passage of time, or addition of chaos, stress, pressure, or unexpected occurrences. Think of a solid piece of metal that will never be exposed to extreme heat.
The Anti-Fragile: These things will grow stronger over time as a result of exposure to chaos, stress, pressure and unexpected occurances up to a point Think of how our muscles grow by creating micro-tears when we lift weights.

Crucially, antifragility means being ok with not understanding the world. That’s key, because most of us get in trouble when we think we can understand everything. The framework assumes that we don’t, and only measures the variables which are clear enough for us to see and measure.

Applied to Foolish Investing

As an investor, I want to put my money into as many “Antifragile” companies as I can. After reading much of Taleb’s work, there are three major concepts that play a role in how do this. In order of importance, they are:

The barbell strategy
In the realm of business, this means having one aspect of your business that is very reliable and accounts for a majority of your time and resources. But on the other end of the “barbell”, you tinker with high risk/high reward experiments.

The first end, I call the MOAT. In my criteria, there are only four main moats a company can have. Here’s how I break that down by points.

Low-Cost Production
+1 point if it would take a competitor up to three years to match this advantage.
+2 points if it would be prohibitively expensive and take (in my best estimation) over three years for a competitor to match.

Example: Think Amazon’s fulfillment network – it can offer one-or-two-day delivery service for less than anyone else can afford, internally.

High Switching Costs
+1 point if these high switching costs are clearly demonstrated, but for less than five years. Or the overall switching costs are somewhat weak.
+2 points if the switching costs have been demonstrated as robust over a long timeline.

Example: I will probably always use Intuit’s TurboTax because it has so much of my prior data in it and it makes filing taxes so much easier for me.

Network Effects
+1 point if the network effects are weak or slowing, or if they are strong but have not been demonstrated over five years.
+2 points if the effects are strong and have been for over five years.

Example: Facebook’s properties get more valuable with the addition of each user.

Intangible Assets
+0 points if a company has a strong brand, but hasn’t proven itself over seven years minimum.
+1 point if a company relies on patents, or has a strong brand with over seven years of proof.
+2 points for regulatory protection of some kind.

Example: Apple has, by far, the most powerful brand in the world.

On the other end of the barbell, we have optionality, or multiple futures. Here’s how that breaks down.

+3 points if the company has shown it can branch into other industries, and has a history of rewarding small, risky failures.
+2 points if the company has shown that it can grow in multiple ways within its current industry.
+1 point if the company has a clear history of trying to branch out into new areas, but has yet to succeed.

Example: Amazon is famous for rewarding small failures, which has led it to go from being an online bookstore to “The Everything Store”

In addition, I like think a company’s mission statement is paramount. It needs to be simple (a compass to help an employee make a difficult decision), inspirational, and optionable. Here’s how that is evaluated:

Mission statement
+2 if it meets all three criteria
+1 if it meets two of three
+0 if it meets one of three
-1 if it meets none.

Skin in the game
Taleb points out that the best way to mitigate hidden risk is to evaluate skin in the game. He points out how, using Hammurabi’s code, ancient cities would require a bridge architect to sleep underneath the bridge after it’s completion. Since the city elders weren’t architects themselves, this was the best way to assure that no corners were cut.

I evaluate this on three different levels

Role of founder
+1 point if founder is still involved in some part of the operations or is on the board.
+0 if he/she is not

Insider holdings
+1 point if the group as a whole owns more than 10% of shares, or one individual owns more than 5%
+0 points – in between the above and below
-1 points if the group as a whole owns less than 5% or there are no individuals owning at least 2%

Glassdoor Employee Reviews
+1 point if the overall score is a 4.0 or higher
+0 points if the overall score is between 3.6 and 3.9
-1 point if the overall score is at or below 3.5

Redundancy and robustness
This is the third major aspect that I evaluate. This is the boring stuff that only becomes important in times of crises. If a company has these characteristics, it is less likely to suffer over time due to unforeseeable stressors.

Financial Fortitude – involves looking as cash, debt, and free cash flow
+1 point if it would clearly benefit – relative to competition – from a downturn
+0 points if it would remain unchanged
-1 point if it would be weakened – relative to competition – in a downturn
-2 points if it might go bankrupt in a downturn.

Customer Concentration
+0 points if it has no concentration issues.
-1 point if top three customer are 15% to 25% of sales
-2 points if top three customers are more than 25% of sales

The big question: Does this work?
I have identified two huge blind spots with the Antifragile framework. First, if a company has none of the four moats I mentioned above, but has incredible business momentum, it can do very well despite the low score. NVIDIA is a great example of this.

Furthermore, because of the binary nature of results in the pharmaceutical industry, it is a poor indicator of success in this realm.

I don’t really view this as a problem, so long as it’s understood, and you decide not to short such stocks.

In terms of the results, I have three data points for you, all fairly encouraging.

My own personal results
I haven’t explicitly used the Antifragile framework from the get-go (I began formulating it about 30 months ago). That said, I’ve unknowingly been using pieces of it throughout.

The results, according to My Scorecard:

Compounded annual growth rate: +20.7% per year
Versus the S&P 500: +5.8 percentage points per year.

That might sound like a small amount of outperformance, but it’s pretty signficant over time.

Cross referencing with a portfolio I started in 2011

When I started at the Motley Fool, I brashly made something called “The World’s Greatest Retirement Portfolio”. You can read about it here:…

Well, I went back and constructed what the Antifragile score would have been at the time, and cross-referenced it with the returns on those stocks over the past six years.

The results were pretty shocking, an r-squared value of .69. In other words, 69% of the variance in returns could be explained by variance in the scores.

That’s pretty high, but it should be said that it also is a very small sample size.

Votes for Explorer Missions
Over a smaller timeframe, I’ve voted on David Gardner’s Explorer missions for the past three years. While I won’t reveal any of the data here, if I would have chosen the “Antifragile” pick every time (I didn’t start using it until 6 months in).

Of the 34 picks I’ve made between two stocks:

The “Antifragile” one has outperformed 17 times (50%)
The average “Antifragile” pick has outperformed the other by 27 percentage points!

I bring all of this to your attention because, inspired by TMFTypeoh, I’m going to be running many stocks in the Motley Fool universe through this framework, and wanted to have a posting to refer back to to explain how certain stocks got their scores.

I hope you find it as educational, amusing, and enrching as I have, and that it helps you to invest – better!

Fool on!
Brian Stoffel

See here for my holdings:


Woops. Failing basic math: Score should be 7.5

Thanks, Brian! 7.5 seems like a pretty good score, and I think a lot of these metrics are things that PSTG is improving. Might be interesting to check back in after a few more quarters. But I think we’re on the same page on this one. It’s no Shopify, but there are reasons to be impressed by what the company is doing and its long term focus.



To me PSTG seems to have two unique aspects, and qualitatively a few.

First, PSTG has a truly sticky business model with the Evergreen. Why the heck would you ever replace your PSTG equipment when you can just upgrade it. Their product is not just gonna be cancelled. Therefore, there is a long-term money stream, that is equal to at least any SaaS company.

Second, if AI and ML and the like becomes the dominant source of data and need for storage in the world, the PSTG is clearly a disruptive force in the industry. No one else has what PSTG offers, and it is unlikely anyone will catch up in regard.

My research has indicated that Network Appliance does produce product that can compete with PSTG on latency, as an example, but not on Evergreen, and they do not have a specific AI focused product (although NTAP can probably, more than any other competitor, come up with something).

Qualitatively, the customer service PSTG provides is unprecedented in an enterprise, hardware type of company (I consider nutanix to be a software company, even when they sell hardware as the wrapper).

The question becomes, just how large of a CAP will this provide PSTG?

My concern is the rapidly slowing growth that PSTG is guiding towards (45% this quarter is great, but 305 thereafter!). I mean that is great long-term growth for any company, but it also is evidence of other things.

What I mean by this, is one has to take into account the reoccurring nature of PSTG’s revenues. Very sticky. And yet with this base, that will continue to grow, revenue growth will only be 30%!

If PSTG was truly disrupting the industry I think it should be growing faster.

It may very well be that the AI market is just not timely yet, and at some time in the future it will tornado and blow all current projections away.

But I am just trying to look at proxy points to figure out PSTG.

We know PSTG is leading market innovation, and mentioned by all its competitors to prove it.
We know PSTG has the best customer service.
We know its Evergreen model is disruptive.
We know PSTG has the best (and really only) AI designed data system.

Innovation is great, but will it translate to disruptive dominance in the industry?

I guess at this point PSTG is that the “not quite there” stage, where its potential looks like it is ready to move to be one of those rare great disruptive companies, but one cannot be sure about it, and the surrounding evidence as to whether or not it can get there is ambiguous.

I do think that the AI market is still pre-chasm, as I like to refer to it, but yes, it seems very likely it will grow very rapidly going forward, and this may account for some of the ambiguity in the proxy data.



Exactly my take. I’d need to get a lot more info. on the specifics before making an investment, but that’s because I’ve just started looking at it.


I don’t know about Pure Storage. Don’t know if their business is awesome or not.

I do know that the meta data being collected by AT&T is growing at a phenomenal rate. What is more, with connected car, Alexa, Siri, and other helper along with every app on my phone wnating more data, there has got to be a place to store it.



Hi Brian, An addition problem. Your numbers on PSTG add up to 7.5, not 6.5. Not sure that makes a lot of difference but you might want to correct it for your own records at least.

Interesting post . . . a few observations:

You stated:
I prefer to buy my companies with a decades long time horizon and – for the most part – following through on that intention.

But then you asserted that you subscribe to the fragility theory which you quoted:
The future is completely unpredictable, and if we use a narrative to try and predict how things will unfold in the world beyond our immediately control, we will likely be wrong.

I find this two statements massively internally inconsistent. They are mutually exclusive.

But, setting that aside, I have some practical concerns:

1) Awfully labor intensive. I don’t know how you value your time, but I wonder if the benefits you get from this analysis are worth the effort. It seems to me that their are easier ways to arrive at similar conclusions. In light of the underlying assumption, the future is completely unpredictable, this is not a do it once, set and forget exercise. On would think you have to crank through this activity every quarter as new information comes to light, maybe more often than that because new information does not always patiently wait for the quarterly report. Seems like an endless, labor intensive task. Now expand that to a large universe of companies - do you ever sleep? Do you have any outside interests? Do you have a family? When do you find time for those things?

2) Much of the information needed to perform this analysis is not easy to come by, in fact I’d venture that some of it comes from an educated guess at best. The old adage GIGO applies irrespective of whether the computer is electronic or biological. “Garbage in, garbage out” - or, slightly modified, “Guesswork in, garbage out.”

But good luck with that.

BTW, have you done a regression testing on this method? Have you looked at companies with high barbell anti-fragile ratings of 20 years past to see if they actually meet your “decades long time horizon” criteria via superior performance as compared to what? Certainly beating the S&P would be a weak criteria. Saul has demonstrated a much simpler path to decades long outstanding performance utilizing a much less labor intensive method. Brings me back to my first point, what’s the payoff for all the input effort?


Sorry, I should have previewed. I meant to close the italics after “…we will likely be wrong.


On your first point: this is just a matter of where I lie on a continuum from “Trading Every Day” on one side, to “Never Selling a Position Ever” on the other. In general, my aim was to say that, here on Saul’s board, positions are moved in and out of on a pretty swift pace. There’s nothing wrong with that, it’s just not how I operate (it would be near impossible with Motley Fool trading rules, anyway). Instead, I’ll make them on a quarterly-ish basis.

As to your other two concerns:

  1. I was motivated to start publishing my findings because TMFTypeoh started doing the same with his own framework. I would say the average run-down take about ten minutes for a company I know well, much longer for the ones I don’t. I’m comfortable with shooting for ten a week – though whether I hit that goal largely depends on what else is going on in my life.

Incidentally, I limit myself to 15 hours of hard writing per week. Beyond that, I’ll read whenever I want, and that can often lead to some interesting ideas. Mostly, I spend time with my wife and 4 year-old daughter.

  1. The info. is actually easier to get to than you might think. Some stuff never or almost never changes (mission statement), Yahoo! Finance is a great place to get balance sheet figures, doing a word search on SEC filings can get much of the rest, Glassdoor is very easy to use. It’s the moat and optionality that really take time to figure out.

As for the payout, all I can tell you is this: over the ten years I’ve been investing, my CAGR is roughly 23%, outpacing the market by 7 percentage points per year. That’s allowed my wife and I to stay at home with our daughter, live part of the year on a coffee farm in Costa Rica, and pretty much volunteer in our community whenever we want.

I really appreciate the question. My guess is that as time goes on (I’m 36), my strategies will become simpler and simpler. I haven’t put in the sweat that Saul has to reach such simplicity, and I’m enjoying the process.



FYI, “Network Appliance” hasn’t been “Network Appliance” for like 9 years now… they changed the name to NetApp in early 2008. I also wouldn’t label PSTG as a “first-mover.” They may have garnered a solid piece of the flash storage arena over the years, but there were many in front of them. I leave that research to anyone who cares.

In other news, today’s whole-market fall might be an opportunity for me to finally own some PSTG… which means all of you should prepare for another drop right after I buy some. :\

1 Like

Congratulations! It’s hard to argue with success.

I’m always stunned by how quickly some folks can gather and employ analytical information. If I recall, Saul says it takes him about 1/2 hour to gather basic fundamental information. You suggest ten new companies/week at 15 hours a week of dedicated effort (you didn’t exactly say that, but you kind of said that). In any case, ten a week is impressive.

I am much slower in the gathering phase, it just takes me a while to figure it out. Some of it I don’t even know how to derive if it’s not reported somewhere (FCF, for example). And analysis takes me much longer as well. I know it’s due to inexperience, I’ve been a serious investor for only two years. I look at a company’s gross margin, for example, it’s x%. Is that good or bad? I don’t know off hand, I don’t have sufficient experience to know immediately just be looking at the number. I lack a comparative framework.

You didn’t address my question about back-testing, but maybe it’s impossible to do. Getting the data from a decade ago would be difficult if not impossible. And if you only look at companies that are successful today with 10 or more years of history it would not be a very representative sample. There might be a ton of companies with great barbell numbers ten years ago that have gone out of business, you’d miss 100% of them.

Anyway, good luck to you, though it appears you don’t need it.

1 Like