Every Business Is a Bet. Size It Like One.

Poker players have an unfair advantage in business, and most of us don't use it.

We already think in bets. We already know that a good decision can lose and a bad one can win. We've all watched someone run hot for a month and decide they're a genius.

So here are the three poker ideas I actually use when I pick what to build. No made-up hero calls, just the concepts.

1. EV: price the whole tree, not the dream

When you're deciding to call, you don't ask "could I win this hand?" You ask what happens across every way the hand plays out, weighted by how often each one happens.

Business ideas deserve the same treatment. The dream outcome is one branch. There's also the branch where nobody wants it, the branch where they want it but won't pay, and the branch where it works but eats all your time.

A hypothetical to make it concrete. Say an idea has a 10% shot at being a real business and a 90% shot at being nothing. If testing it costs you two weeks and a few hundred bucks, that can be a great bet. If testing it costs six months and your sanity, the same 10% is a terrible one.

Same odds. The price changes everything.

Last year I wrote something I still believe: "EV doesn't matter if you don't have the hours to capture it." Your time is part of the price. Most people forget to put it in the pot.

2. Variance: decide what "it didn't work" means before you start

In poker you can play perfectly for a week and lose. That's variance. The answer isn't to change your strategy every time you lose a pot. You set a sample size you trust and judge yourself on that.

New businesses have brutal variance. A good idea gets zero sign-ups the first week because you posted it on a Tuesday at 3am. A bad idea gets a spike because one person with a big following was bored.

So every company in my portfolio now gets its test written down before it starts:

• Snagworth gets 28 days. On day 28 it's continue, pivot or kill, against a bar written on day 0. Its charter literally says "The clock is fixed." No seat is allowed to ask for more time.
• SlotMaps has a mandate: at least one new paying subscriber every month from October through March.

That's not pessimism. It's the same thing as not chasing losses. You decide the stop-loss while you're calm, not at 4am when you're stuck three buy-ins.

3. Bankroll: never put the whole roll on one table

Bankroll management is the most boring idea in poker and the reason good players survive. You don't sit in a game where one bad night busts you.

For a business, your bankroll is money and hours and attention. I've overdrawn all three. My "Starting Too Many Things™" disease is a bankroll problem: too many tables open at once, none of them getting my A-game.

What I do about it now:

• A fixed budget per experiment. When I gave an AI agent $1,000 to build a business in March, that $1,000 was the whole roll for that game. It never got a reload.
• A fixed time box. See variance, above.
• Delegate the grinding. The thing that changed my bankroll math the most was AI seats doing the work that used to eat my hours. More tables open, without me in every seat.

The quiet one: table selection

Bonus round. The best players I know win as much by choosing the game as by playing it well. A great player in a tough game can make less than an average player in a soft one.

Markets work the same way. Before you ask "can I build this?", ask "who else is sitting here, and how good are they?" A market full of slow, expensive incumbents is a soft table.

Try it on your next idea

Before you build anything this week, write four lines:

1. The tree: best case, likely case, nothing case, and a rough % for each.
2. The price: dollars and hours to get a real answer.
3. The clock: the date you decide, and what number means "keep going."
4. The table: who's already in this game, and are they good?

If you can't fill in line 3, you're not testing a business. You're just gambling without a stop-loss. And I say that as a gambler.

Post your four lines in the community. I want to see them.

— Art