The ROI Ceiling Nobody Knows About

Bob had been lurking in a forum thread titled "What ROI is realistic for MTTs?" Someone claimed 80% ROI in $55 turbos. Someone else called bullshit. Fourteen pages of graphs, screenshots, one guy called a donkey. Bob screenshotted it and sent it to Tau. Tau replied with a single number and stopped picking up.

Bob: You sent me a number and then ghosted.

Uncle Tau: It's the ceiling. The maximum ROI a player can achieve in a standard 200-runner online tournament at a typical cashing rate. Not an estimate. Not opinion. It's computed.

Bob: People in this thread are claiming triple that.

Uncle Tau: They're wrong, or they're in fields soft enough to bend the assumptions, or they ran hot for a month. At a normal rate in a standard payout structure, that number is where the math stops them.

Bob: How can ROI have a ceiling? If you're better, you win more.

Uncle Tau: "Better" means finishing higher more often. Finishing higher means beating people. Beating people has a bound. Even the best heads-up player alive doesn't sustain more than about sixty percent of heads-up confrontations against comparable opponents, long-term. That bound propagates outward. It limits how much mass any honest forecast can push toward the top of the finishing distribution, which limits the payout you can expect, which limits ROI.

Bob: And that's computable.

Uncle Tau: Given a payout table and a heads-up cap, yes. There's a theorem. The ceiling is the ROI at which the math refuses to budge any further. Short proof, long proof, both written up. Neither lives on this page.

Bob: You're doing the "trust me, there's a theorem" thing again.

Uncle Tau: I am. The closed form is behind a subscriber call, the same way SALSA is. The framework is the edge. Once the framework is out, the edge isn't.

The counterintuitive part

Bob: Give me one number. You can spare one.

Uncle Tau: Fine. The 200-runner you asked about, the one the forum is fighting over — the ceiling is in the high twenties. WSOP Main Event, with ten thousand runners and a thousand-buyin first prize? Lower than the 200-runner.

Bob: That's the opposite of what I'd guess.

Uncle Tau: It's the opposite of what everyone guesses. The size of first place doesn't matter the way your instinct says it does. What matters is something about the shape of the payout distribution — something I'm not going to walk you through on a free page. The app knows the shape. The app uses the shape. That's the product.

Bob: Payout structure sets the physics, field determines where I sit inside the physics.

Uncle Tau: That's the line. Keep it.

Back to the forum thread

Bob: The guy claiming 80% in 200-runner $55s.

Uncle Tau: Not impossible. But it requires him to be near-ceiling on more than one thing simultaneously — and nothing in a forum post tells you whether his data supports that. The argument is over before it starts because neither he nor the guy calling him a donkey has computed where the ceiling is for his structure. They're debating the location of a wall they can't see.

Bob: Why doesn't everyone just compute this?

Uncle Tau: Because the math lives in a different field from poker. Tournament lifers don't read what this draws from. The connection isn't obvious until someone shows you — and most people in poker haven't been shown, and aren't going to be shown for free.

What to do with this

Bob: Practical.

Uncle Tau: Two things. One: your ROI has a ceiling, and it's specific to the tournaments you play and your realistic win-rate against the field you actually face. If you're claiming numbers above what the structure allows somebody with your profile, you're wrong in a way that's measurable. The app does the measuring. Point it at your data, it tells you where you sit relative to the ceiling.

Two: when you hear people argue about "realistic ROI" on forums or Discords, understand they're arguing about a number that's computable, not a number that's open to opinion. Where inside the feasible space a specific player sits is empirical. That the space has hard edges is not.

Bob: Five sentences and a ceiling.

Uncle Tau: And fourteen pages of forum noise it kills.

Go estimate your shapes, kid.

What's next

• Sample size — how many tournaments is "enough"? — why some skill parameters converge in hundreds of tournaments and others take ten thousand, and what that means for staking decisions.
• Confidence interval width — how to read the width of every posterior in the app, and when to act on a number versus wait for more data.

Further reading

• The closed form, the derivation, and the full numerical tables for every structure are in an internal paper. For depth — that's a subscriber call, not a wiki page.
• Priors and Posteriors — why every number in this app has a width.
• SALSA in one sitting — the simulator that converts skill into a finishing distribution.