To The Moon

$AppLovin Corporation(APP)$  

Mathematical Money | October 2026

Run AppLovin's valuation backwards and you get a precise answer. At Tuesday's close of $262.19, the price assumes free cash flow grows 20% a year for a decade. After that, 2.5% forever.

The company reported revenue growth of 52.8% last quarter.

A gap that size normally means something is badly mispriced. I own this one, I have been wrong on it all year, and I am still not adding — because three other numbers from the same analysis say the obvious conclusion is a trap.

Here is the whole thing, including the checks you can run yourself.

Check one: what growth is the price already assuming?

Most people value a company forwards. You forecast growth, discount it, get a number, compare it to the price. Every input is a guess, so the output is a guess with decimal places.

Run it backwards instead. Take today's price as given and ask what growth rate makes it fair. That single question strips out your own optimism, because the market has already supplied the answer and you are only reading it off.

How to run it on your own holdings. You do not need to build anything, and several free calculators will do it. Take the company's free cash flow, its price, and a discount rate somewhere around 10 to 15%. Then ask one question: at what annual growth does this price make sense over ten years? You are not after precision. You want to know roughly where it lands — five percent, twenty, or forty-five. That is the number you are actually betting against.

AppLovin's answer is 20.0% a year for ten years. After that, 2.5% forever, discounted at 15%.

What it tells you. The market is not pricing a deceleration from 53% to 40%. It is pricing a company that grows at 20%. That is a different business entirely. Either the market is wrong, or it knows the 53% does not survive contact with next year.

That is the bull case in one line, and it is the reason I own it.

Now the three reasons I am not buying more.

Reason one: the earnings are flattering the business

This is the check almost nobody runs, and it is the one that stops you buying things that look cheap.

Compare a company's current operating margin to its own mid-cycle average. Not to rivals — to its own history. If today's margin sits well above that average, today's earnings are unusually good, and every multiple you calculate off them is unusually low.

AppLovin's operating margin right now is 75.8%. Its mid-cycle average is 43.8%.

So the forward multiple of about 13 times earnings is not describing a cheap company. It is describing a company earning at the top of its own range. Normalise the margin and the multiple roughly doubles. A trailing multiple understates how expensive this is, which is the exact opposite of what the headline number suggests.

Run this on anything you own that screens cheap. Find its operating margin today, find the average across the last five years, and compare them. If today sits well above the average, you are not looking at a bargain, you are looking at a peak. It is the fastest way I know to tell those two apart.

Reason two: our own model refuses to give a number

Four independent methods, run on the same filed statements:

Earnings power says $211.82.

Book value says $141.51.

A cash-profit multiple says $304.73.

A two-stage discounted cash flow says $364.12.

The spread from lowest to highest is 157%. Our framework's rule is that when methods disagree by more than a threshold, it does not publish a point estimate at all. It hands back the range and tells you to use it.

I find that more useful than a target price, and I want to be explicit about why. Any research note that gives you one number for this company is hiding the disagreement, not resolving it. Four reasonable methods land between $142 and $364. The honest answer is that the fair value of this business is genuinely unknown right now, and anyone certain about it is selling you something.

Reason three — and this is the one that decided it

I hold long-dated calls struck at $220. With the premium paid on top, they break even at $365.01.

Now pull the chart's own resistance levels, measured from pivot points rather than drawn by hand:

$300.25, touched 3 times, most recently 24 days ago.

$363.41, touched 5 times, most recently 129 days ago.

$434.02, touched 2 times.

The level this stock has been rejected at five times sits $1.60 below where my position starts making money.

That is not a coincidence I enjoyed finding. To make money on these calls, the stock has to clear the single most defended price on its own chart and keep going. Not reach it. Clear it.

Here is the check, and it costs you nothing. Put your break-even on the chart next to the resistance levels. If the wall is above your break-even, the position has room. If the wall sits on top of it, you need the stock to do the hardest thing it can do just to get you to zero. Mine does.

What is genuinely good, because it is not a bad company

The quality numbers are not the problem, and I want to be fair to the business:

It scores nine out of nine on the standard nine-point test of whether fundamentals are improving. The bankruptcy-risk score sits deep in the safe zone at 13.80. Return on invested capital runs at 63.9% against a cost of capital of 15% — a spread of nearly 49 points, meaning growth genuinely creates value rather than consuming it. Incremental returns on new capital are 231%. The share count is shrinking 8% a year on buybacks. Free cash flow is $3.94 billion, and holds at $3.73 billion even after stripping out share-based pay.

One detail I liked: receivables are growing 28 points slower than revenue. When a company inflates sales by shipping to customers who have not paid, receivables outrun revenue. The opposite is happening here. The 53% is real cash, not an accounting trick.

And the shape of the growth, which is the fact I keep returning to: revenue per app install rose 58% while install volume fell 2%. All of the growth is price. The engine is extracting more from the same inventory rather than reaching more people. That is either a brilliant AI pricing engine or a fixed pond being repriced, and the honest answer is that nobody outside the company knows yet.

What actually broke the share price

Not demand. A disclosure.

On the 5 August call, AppLovin said its flagship AI video creative tool remains a work in progress. It cannot yet deliver high-quality 30 to 60 second videos to advertisers out of the box. Earlier guidance had said the feature would roll out to all accounts shortly. A 13 July analyst note had pre-warned it. On 16 September a securities class action was filed in Northern California alleging the company misrepresented its AI progress, covering February to August 2026. The lead plaintiff deadline is 16 November.

So the open question is not advertising demand. It is whether the AI story was oversold, and what that costs in court. A litigation discount of 15 to 20% is reasonable until a settlement range is visible.

The twelve-month view, stated as a range because that is what it is

Rebased to Tuesday's close:

Bull, $420 to $550. Midpoint $485, probability 30%. Needs the AI tool at production quality and 2027 guidance raised.

Base, $320 to $420. Midpoint $370, probability 50%. Growth settling at 30–40%, the class action settling cheaply.

Bear, $180 to $320. Midpoint $250, probability 20%. A third-quarter miss, the AI tool abandoned, a material settlement.

Probability-weighted, about +45% over twelve months. The analyst median is $475 across 33 analysts. Jefferies cut to $375 on 9 October, and revisions have been heading down since August.

Note what that does to my calls. The base case midpoint of $370 clears my $365 break-even by five dollars. Half the probability mass lands on being right and making almost nothing.

Where the price sits, measured rather than eyeballed

It trades 40% below its 200-day average and 15% below the 50-day, with both of those averages falling. On the standard four-stage trend framework this is stage four — decline, which is explicitly the stage to avoid committing new capital in.

It is 0.5% off its 52-week low and just made a new 55-day low. Momentum is deeply oversold at 30.6, which often precedes a bounce — but the trend behind it is still down, and relative strength against the S&P has fallen 48.7% over six months.

One thing on the other side. It still sits 7.5% above its 200-week average of $243.81. The multi-year trend has not broken, only the one-year trend has. That is the line I am actually watching, and it is a long way below here.

Realised volatility is running at 48% and the options are priced for more than that, which makes this a far better name to sell options against than to buy them. That is the only part of my position currently working.

The dated tests

4 November, after the close. Third-quarter results. Not revenue first — I want the AI video tool status and a specific production timeline.

16 November. Lead plaintiff deadline on the class action.

The arithmetic says the market is paying for a 20% grower and getting a 53% one. I own it because of that. I am not adding because the margin is at the top of its own cycle, our own methods disagree by 157%, and the wall on the chart sits $1.60 under my break-even.

Those three checks took about twenty minutes between them, and you can run all of them on your own holdings this week. Work the price backwards and see what growth it already assumes. Check whether the margin is flattering the multiple. Put your break-even on the chart next to the walls.

If you only do one, do the first. It is the fastest way I know to find out whether you are buying a business or a story about one.

#APP #valuation #investing #stockanalysis #options

Stop guessing. Start calculating.

Live to fight another day. 🤙