A daily board that prices every MLB game against the book — and says so when there's nothing worth taking. College football and the NFL next.
It started as a group chat problem.
My friends bet on MLB every day. They weren't asking for more data — they wanted someone to just tell them which bets were worth taking.
They kept bugging me about it. So I built it.
Hours of research still don't tell you if a bet is worth taking.
You can read every stat on a game and still be guessing, because a sportsbook never shows you the one number that matters.
Do the work for them — then show every step.
A number on every game, not just the ones we like
Every game on the slate gets our number next to the book's, ranked by the gap between them. Anything clearing the bar carries a quarter-Kelly stake. Most days that's one or two. Often it's none, and the board says so rather than filling the space.
Every number says why
Every row opens into plain sentences saying why — the starters, the park, what the model checked and ignored. The full reasoning sits behind a toggle for anyone who wants to check that it adds up.
An hour of scrolling odds became thirty seconds, with the reasoning attached.
Loading the live app…It answered the question early. Then it kept me honest.
My friends check it before first pitch instead of asking each other. Every pick gets graded off the final score, and I've never edited the record.
A win-loss record needs around 800 games to mean anything. Closing-line value — whether our number beats the price we bet into — needs about a hundred. Both tests were written down before the data existed, including the one I ran when a pattern turned up in the losses: every road pick, 42%, at every price. Neither cleared. Nothing changed.
Where it goes from here.
Four times the games, so it validates in one season instead of three.
Every instrument the model is judged by↗, including the ones that came back flat.
Every new signal gets a bar written down before it's measured. Five have failed that way.
Write the bar for charging while there's no money on the table, not on a hot streak.
Anyone can build a model that looks right. The edge is being able to prove whether it is — and doing what the answer says.