Kevin Warsh spent his Jackson Hole keynote refusing to tell anybody what the Fed is going to do.
What he did say is that inflation is running above target, that he treats 2% as a firm and fixed number, and that short-term rates are the predominant tool for getting there. The market moved the odds of a hike this year to roughly two in three.
So we walk into September, historically the worst month of the year for stocks, with a Fed chair who sounds like he wants to raise and will not say when.
I have been getting the same question all week. Where do you put money when you cannot read the next six months?
The old answer was that you buy businesses nobody can compete with. Wide moats, meaning Coca-Cola and Visa and the railroads, the names your grandfather owned.
That answer is still mostly right, and I want to be careful how I say this, because something changed underneath it and almost nobody has adjusted.
A moat only counts if the thing crossing it cannot fly.
Below I will give you the whole list, the three numbers that define a great business, what these names did in 2022 when the market came apart, and which of them have a moat that a machine can now walk straight across.
Start with the three numbers, because the word moat gets thrown around by people who cannot define it.
A great business has three things. Large free cash flow, low capital requirements, and a high return on shareholder equity.
Large free cash flow means the money shows up in the bank instead of sitting in a warehouse as inventory. Low capital requirements mean the business does not need to keep spending enormous sums just to stand still. High return on equity means every dollar the owners leave in the business comes back working.
Get all three and you have a machine that prints money without needing to be fed. Visa and Mastercard score a perfect ten on all three. A railroad scores a seven on cash flow and a four on capital requirements, because track is expensive and it never stops needing maintenance.
The moat is the separate question. It is the structural reason a competitor cannot show up and take those returns away.
There are five that hold up. Switching costs, where leaving is more painful than staying. Network effects, where the thing gets more valuable as more people use it. Cost advantages from scale that a smaller rival cannot match. Intangibles, meaning brands and patents and licences. And efficient scale, where the market only supports one or two players profitably.
If you cannot name which of those five a company has, then it does not have a moat, it just had a good quarter.
What that bought you in 2022
Here is where the argument earns its keep.
In 2022 the S&P 500 dropped 24.5%, the Nasdaq dropped 33%, and the Dow gave back 20%.
Coca-Cola fell 7%.
Stretch it out and it gets better. Put $100 into Coke at the end of 2017 and by the end of 2022 you had $163. The same $100 in the S&P gave you $157. The moat name won that whole stretch because of the one year everything else broke.
These names never shoot the lights out, they just refuse to fall apart.
Now the part nobody mentions
Over the last ten years the VanEck Wide Moat ETF returned 13.90% a year. The S&P 500 returned 15.34%.
You gave up a point and a half a year for a decade, because moat screens systematically underweight the mega-cap growth that drove the entire period.
So be honest about the trade. You are giving up the top end in exchange for a much narrower spread of outcomes. In 2022 that was the best deal in the market. From 2023 through 2025 it was a bad one.
The new test
Every one of those five moat types was defined before machines could write, design, analyse and advise.
So there is now a sixth question, and it goes in front of the other five. Can the moat survive artificial intelligence?
Run the list through that and it reshuffles badly.
Alphabet’s moat is search dominance, and search is the single most obviously disrupted thing on earth right now. Microsoft’s Office franchise faces AI-native alternatives. Adobe sells creative software into a world where the software generates the creative. Intuit sells tax and accounting automation to a market that AI can automate more cheaply. Booking sells travel planning, Accenture sells consultants, and TransUnion sells credit scoring built on static models.
Every one of those is a genuine moat by the old definition. Every one of them is a moat that something can now fly over.
What AI cannot cross
Look at what survives the same test.
Visa and Mastercard, because a payment network is infrastructure and AI does not settle transactions.
Coca-Cola, because a machine cannot manufacture a hundred years of brand and a bottling network.
Johnson and Johnson, because physical medicine gets made in factories and approved by regulators.
Walmart and Costco, because AI improves their logistics without replacing their stores.
LVMH, Hermès and Ferrari, because those businesses sell scarcity and craftsmanship, and a machine that can make more of something is the enemy of exclusivity, not the tool for it.
Waste Management and American Water Works, because somebody still has to move the garbage and treat the water.
Canadian National and Union Pacific, because you cannot generate a railroad.
Now hold both lists next to each other and look at what falls out.
The paradox
The businesses with the best financial profiles are frequently the most exposed.
Asset-light, high return on equity, low capital requirement. That describes software and data and ratings and consulting, which is almost everything on the vulnerable list.
Meanwhile the names that score worst on capital requirements are the safest. NextEra scores a three, and American Water Works, Walmart, Canadian National and Union Pacific all score a four. Track and pipe and landfill are miserable businesses to fund and impossible businesses to disrupt with a language model.
The expensive infrastructure that made those companies look mediocre for twenty years is now the thing protecting them.
Which leaves one exception worth writing down on its own. Visa and Mastercard score a perfect ten across all three properties and they sit on the resistant side of the AI question.
The trap
None of this tells you what to pay, and that is where people get hurt in exactly this environment.
Right now the list is split down the middle. Coca-Cola is up around 31% this year and sitting at an all-time high. Costco trades near 49 times forward earnings. Both are magnificent and neither is a bargain.
At the other end, Nike is down 39.7% year to date, 50.8% over twelve months, and 77% from its peak.
So Nike is cheap, right? It trades at roughly 22.8 times next year’s earnings.
A drawdown is not a valuation. Price came down and earnings came down with it, and in Nike’s case the earnings dropped faster. Revenue was flat at $46 billion and net income fell 3%.
The stock got much cheaper looking and barely cheaper.
People make that exact mistake when they get scared. They go looking for quality, they screen for the biggest declines, and they end up buying broken earnings at a full multiple.
Compare it to Estée Lauder, which is down 68.8% over five years and just beat its quarter by 40% and raised guidance for next year. Nike’s management has said publicly that the fix will take longer.
Both have real moats, but one has produced evidence and the other is asking you to keep believing.
Which account this goes in
Most traders I know run two, and they run them for different reasons.
The trading account is where the income comes from. Defined risk, short duration, and you are getting paid to carry volatility somebody else does not want. That account does not care what the Fed does in six months because nothing in it lives that long.
The long-term account is where the profits go, and its job is to compound while you sleep. You are not trading it.
Moat names belong in the second account because of one main reason. They make terrible trading vehicles, since they do not move enough to pay for the option premium.
What I would do with the list
Take the whole list and mark each name three ways.
Which of the five moats does it have. Can a machine cross that moat. And what would you pay for it.
The first two you can answer today, sitting down, with no market data at all. The third one the market has to hand you, and it will, usually when everybody else is frightened.
So build the list while you are calm and buy it when you are not.
Andrew Giovinazzi
Option Pit