How to Avoid Bad Stock Predictions (Reverse DCF Explained)

Get smarter on investing, business, and personal finance in 5 minutes.

This week's newsletter is an adaptation from my recent YouTube video on The Reverse DCF.

This is one of my most requested posts so far, and today I give it all away.

We talk about what the Reverse DCF is, how I do it, and I show you with a real example from an investment report I wrote in January 2023 on Meta, plus a couple other stocks.

One of the most common questions you get when you do a discounted cash flow is that it's way too theoretical: “Garbage in, garbage out, no way it could produce anything valuable.”

What I'd say to this is that a DCF is a tool, and there are all sorts of ways you could misuse a tool.

Do not blame the tool if you choose to misuse it.

Rather, learn to use it correctly, and decide if it's helpful to your process or not.

To the people that say it's garbage in, garbage out, why don't we try to be thoughtful in how we construct it and see if it tells us anything useful?

I also never just do the Reverse DCF.

It helps you hone in on assumptions and what is being priced into the market, and gives you an idea of returns, but these are very long-term returns, so it makes sense to look at a stock in more than one way.

How the Reverse DCF Works.

The Reverse DCF is different from a regular discounted cash flow in a simple way.

Instead of plugging in a discount rate and spitting out a market value, we set our entire DCF equal to the existing market valuation.

In this example, pulling from my Meta research report back in 2023, if we look at Meta at a $277 share price.

The enterprise value at the time was $700bn.

You go to Data, then click on What If Analysis, then click on Goal Seek, set the cell that is the sum of the DCF to the current market value, $701.925bn, and get there by changing the discount rate cell.

It solves it for us, and that's where 11.2% came from.

You set the entire DCF equal to the enterprise value, and you get a discount rate as the output, which assumes all of the assumptions embedded in the DCF.

Building the Meta Reverse DCF.

Let me go one layer more complex.

There are usually two variables I sensitize around in a DCF.

You could do more, but it gets complicated, so for Meta it was revenue growth and margin.

I never run this with just one scenario, because you want to see how returns vary.

At the time, people were pretty bearish on Meta, so I threw in a negative 5% scenario and a 0% growth scenario.

Then there's an 8% scenario, where growth holds at 8% for 5 years and then fades to 7%, 5%, and 3%, because I don't want to assume 8% growth for the entire life of the company.

Then you make a margin assumption.

I want to give you permission to build your DCF however makes the most sense to you, because company disclosures are different, and that feeds into how you look at the business.

For Meta, we had Family of Apps EBIT margins, since they disclose that, but there wasn't Capex directed just to Family of Apps versus Reality Labs, their other big Capex spend at the time.

We were living in a quaint world then, when people were freaking out about a $32 billion Capex spend.

I went from Family of Apps EBIT margins to EBITDA margins, so I could attribute all of the Capex to that segment, a more conservative way to do it.

Depreciation and amortization is a cost allocation tool that should match capital expenditures to the lifetime over which that Capex is used, so $100bn in Capex over a 5-year lifetime gets you $20bn in depreciation.

Since Capex was increasing a lot, there was a criticism that D&A was understated.

To skirt that debate, I assumed the full Capex burden on Family of Apps earnings, $47bn of EBIT for the last 12 months, an implied margin of 41%.

Adding back assumed D&A gets you to $56bn.

Reality Labs was burning a lot of money and not making much, so I added another layer of conservatism: I assumed they burned another $57bn in earnings over about 5 years, with almost no revenue, then shut it down.

I'm writing research, and I have to decide which debates I want to take.

If a stock isn't priced very expensively, you can burden it with a lot of things people think are wrong in the business and sidestep that whole debate.

I didn't want to argue whether Reality Labs would be successful.

Instead, I assumed it was fully loss-making and looked at the returns.

We subtract out Capex, then back out tax, and that gets us an owner's earnings figure.

Owner's Earnings vs. Free Cash Flow.

Owner's earnings are the earnings a business is really generating, the ones you could think of as belonging to the business.

They don't have to track cash flow 1:1, since they ignore a lot of operating working cash flow items, but over the longer term they track pretty closely.

You are welcome to just discount free cash flow, though technically you shouldn't burden it with all of the Capex, only the maintenance Capex.

If you look at Meta today, operating cash flow is $130bn and Capex is $90bn, so that's $40bn of free cash flow, and if you back out stock-based comp, FCF is really $15bn.

That's not a lot of cash flow.

If you tell me Meta is only generating $15bn in free cash flow on an ongoing basis, something is wrong with that number, because a lot of the capex they're currently using is to generate future returns.

That's also why you can't put a multiple on a depressed number like that, and why you back out the growth Capex to get a normalized figure.

For a DCF, because you model every period, you can throw in all of the Capex initially and taper it over time, but then you need to show future periods of higher growth.

Otherwise you're implying the return on invested capital of that growth Capex was minimal or negative.

Running the Scenarios.

Let's do the 5% growth scenario.

In 2023, EBITDA margins were depressed on Family of Apps, so I used $56bn less $18bn, which gets us a 38% EBIT margin.

Since we jacked down the growth a lot, the DCF responded very poorly.

We run the What If Analysis again, set equal to the current enterprise value, and the return drops to 8.5%.

That means 8.5% is the return you'd get if you owned the entire business and they grew 5% for 10 years, then 4%, then 3%, with those margins.

This works because we're solving for the current market price.

I like to keep playing with it, and when you get to around a 9% to 10% return, because that's roughly what the market has historically earned, you could say that's what is being priced in.

At the time of writing the report back in 2023, in the 10% scenario with some EBIT margin compression, the market was pricing in Meta growing basically 3%.

Even in the 0% growth scenario, you're getting around 9% to 10%.

So at the time, at 0% growth, you're getting a ~9% return.

The market was implying that Meta was done growing forever.

That is what the Reverse DCF could show you.

Why These Returns Look Wrong.

You're going to get a fair criticism that these returns look wrong.

Meta was at $130 when I wrote this, and today it's $744.

But this is a return on the lifetime of the business.

If you bought in at that time, that's your return assuming this revenue growth scenario happened and they returned all cash through buybacks, a bit of an unrealistic assumption, because if a company isn't growing, it's probably going to do stupid things with its cash flows other than return them to shareholders.

Very rarely does a business gracefully age.

The more hiccups a business goes through, the more aggressive it gets in trying to find new growth opportunities.

Most CEOs aren't happy being a good steward of capital in a dying business, so they tend to do very risky things.

I'm under no illusions that these are correct numbers, it's to get you a feel for the valuation and price.

At the time, people were really scared about owning Meta even at $130.

If you saw that with just 3% growth you'd get a return above the market average, and that a return to double-digit growth would get you a high teens return, that's a pretty strong picture.

It ultimately comes down to your judgment.

These are very long-term figures, a downside of this model, but it grounds you in the scenarios being priced in, roughly speaking.

Look at It More Than One Way.

Don't look at it just one way.

Here I had a table looking three years out at free cash flow under 3%, 5%, and 8% revenue growth assumptions.

The range of FCF was because of Capex, which we didn't know.

At 5% growth, that's 12-16x free cash flow three years out, fully burdening the business with all of the Capex and giving zero credit to Reality Labs, assuming a negative valuation dragging down the business.

You could take an opinion on that: "Do I want to own Meta where, in three years, that's the free cash flow multiple I get?"

When the Numbers Fool You.

Here is where it can go awry.

When you get too into numbers, it's very easy to fool yourself and get any result you want.

This tool is not going to help you pick better stocks.

In fact, it will make you a worse investor if you misuse it, because studies show that when you give people more information and evidence, they double down on their preconceived notions, and I'm giving you a lot of information you could go astray with.

The inoculation comes down to judgment.

Set it up, then step back and think about what you believe is actually going to happen in this business.

Do I think Meta is done growing forever?

Is a return to growth plausible?

You can think about these returns as a distribution curve.

Everything in investing exists as a probabilistic distribution, which is why I hate when people do a DCF and say, here's a target stock price.

There are different things that can happen, some in Meta's control and others not.

As you learn more business history and watch more deep dives, you'll have more context to make these assumptions.

When you're just starting out, you don't know whether 10% growth is hard for a company.

It's actually really hard for a company to consistently grow 10% for a long time.

Let me make this clear with Etsy.

Etsy Mistake.

At the time they really weren't growing, and there were signs they were shrinking, having a hard time maintaining their COVID gains.

You'd have seen they had to re-accelerate growth just to get a market return of roughly 8% to 11%.

Under the aggressive assumptions of 15%, 10%, 7%, and 3% revenue growth, which look pretty wrong in retrospect, we got a 12.8% discount rate.

There were reasons to doubt they could accelerate.

You can see directionally how this is helpful, at least I think so.

The Reverse DCF number is the sum of all cash flows the business will generate over its lifetime, discounted back to today, where the discount rate is solved from the existing market price.

If you believe the revenue growth assumption and the margin assumption, that's what's being priced in.

But when I did this 79 page report on them, what we saw was a frequency problem: they grew users quickly through COVID, and then it was flat.

I compared it to eBay, which also had a lot of growth and ended up losing it, active buyers fading and fading.

Etsy's buyers were pretty stagnant, so they were holding onto them, and there were proof points that frequency was growing amongst their habitual buyers, the ones most loyal to the platform.

Etsy has always had an issue where, on average, people buy on the platform once every two years.

They were doing a lot around AI and ML, even back then, to surface better recommendations and create less friction, and the thinking was that if it were easier to find what people were looking for, frequency would continue to increase.

Josh Silverman, who ran the business for several years, said people thought of Etsy as a homemade platform, which was a problem because there's no purchase occasion for homemade.

The purchase occasion is special, so whenever you want something special, you think of Etsy.

Margins dipped from 25% to around 20%, but if they had fruitful efforts from S&M and R&D spend, you could get back to a 25% margin.

It's a marketplace, so it should be pretty profitable, and it dropped to just 19x cash flow.

I was never that enamored with the business, and I knew these DCF outputs weren't stellar, but I also thought, how hard is it really to get to high single digit growth?

Ultimately, I owned a very small position for a small period of time.

The reason why I decided to take a small position was a mix of a few psychological biases.

The first was Commitment Bias.

All this research I did: how could I not take a position?

I also felt like I understood the business better than most people, and I thought the frequency improvement could be an inflection, so I was forcing it a little bit.

The second was My-Side Bias.

When something is yours, you're less likely to evaluate it agnostically.

The last bias was Sunk Cost Fallacy.

You put a lot of effort into something, and it's hard to just walk away.

It was overall a good experience, because I realized how I could fool myself with numbers.

When I do a lot of research on a name, I can't fall in love with it.

People say you should learn from other people's mistakes, but I've met very few people who actually do; most good investors have to make their own mistakes to learn from them.

I kind of knew it was a forced position, and it always made me a little uneasy.

The Temu Comment.

I remember Josh Silverman commenting that Temu's ad spend was making it harder for them to acquire users, because it was increasing their customer acquisition price.

When he made that comment, I knew something was very wrong with the business.

Under the consumer hierarchy of preferences, a good Etsy customer can't have a preference for fast shipping, and can't be very price sensitive; they should prefer something handmade and special, not the same mass-produced stuff out of Asia.

If Silverman is alluding to Etsy's customers being indifferent between Etsy and Temu, the antithesis of Etsy, cheap and mass-produced, that is extremely problematic for the marketplace.

As soon as I heard that, I got out of the stock, right around the price I started acquiring it at.

It was fine from a financial perspective, but from a logical perspective and being intellectually honest, it was a disaster, because this is how you fool yourself into an investment with numbers.

The Reverse DCF Is a Compass, Not a Decision.

The reason I'm telling you all this is that a lot of you are going to see a tool that gives you very clean numbers and say, they only need to do that and it's that.

Even if it's only an 8% scenario, is a 9% return that bad?

That's what the market does anyway, and I don't want you to use the tool like that.

It should never come as a rationalization to making an investment.

Think of it as a compass.

It tells you what direction an investment is facing, but it does not tell you whether you should go in that direction.

These output tables show a range, maybe 9% to 18% for Meta, a probability distribution you can think about, but a decision should not fall from it in and of itself.

It should come from your understanding of the business, your comfort with the business, and your comfort with the assumptions and what that means in terms of a valuation.

Valuation vs Pricing.

As I noted, I suggest you look at it more than one way.

A lot of times people think it's a bad word to price stocks, because it gets intertwined with speculating or gambling.

Let's be a little more grounded: anytime you decide what you are comfortable paying for a stock, that's a valuation.

A pricing is what you think other people will end up being willing to pay for the stock.

I suggest you look at it both ways.

The valuation aspect is very much the Reverse DCF: this is what is being priced in, and it gives me a sense of whether I'm willing to accept this return.

The pricing aspect can be a lot simpler, because people aren't going to run Reverse DCFs.

You could ask whether someone would be willing to pay a 25x multiple for a company like Meta if they return to low double-digit growth.

Maybe, or maybe 20x, to be even more conservative: 20x earnings three years out, compared to maybe around 15x, that's 33% upside over three years.

Just because someone is willing to pay a crazy price for a stock doesn't mean someone will pay an even crazier price in the future.

Instead, ask what a reasonable investor with reasonable expectations would be willing to pay for this stock in the future.

That is the Reverse DCF, a lecture on valuation.

I hope the takeaway you get from this is that a Reverse DCF allows you to find what’s priced in given a degree of assumptions.

Think of the table of returns that you get from the Reverse DCF as a probability distribution of outcomes.

Your judgement on the business is key when ultimately figuring out what assumptions you are comfortable with.

For More on The Reverse DCF, check out the video below.

YouTube FAQ

Nothing in this newsletter is investment advice nor should be construed as such. Contributors to the newsletter may own securities discussed. Furthermore, accounts contributors advise on may also have positions in companies discussed. Please see our full disclaimers here.