About MatchWiz

Why we built it

The idea came from a frustration most number-driven people eventually hit: confidence is free, but a graded track record costs something. Any tout can claim their plays are hitting 67%. Almost none will show you the full ledger — every call, every result, dated and auditable. When we looked for that, we couldn't find it. So we built it.

The model came first. The grading layer is the product. A model without accountability is just trivia — it only means something if you can watch it lose and see it learn. That's the whole design: one consistent methodology, applied to every player on every slate, with the receipts sitting right next to it.

How the model scores a matchup

For a hitter, the score starts with the matchup in front of him. How has he actually hit against the kind of arm he's facing? Handedness matters a lot — a lefty's splits against righties aren't the same as his splits against lefties, and we keep them separate. That production goes against what the pitcher and the bullpen behind him typically give up, then adjusts for expected plate appearances, the ballpark, the day's weather, and how the hitter has swung the bat lately. Pitcher boards run the same logic in reverse: the starter against the specific lineup he's drawing. Every player on the slate ends up with one comparable projection.

A projection on its own is just a number. It becomes meaningful when it meets a price. So wherever a book posts a line, we set our projection beside it and measure the gap. The spots where that disagreement is biggest and best-supported become Plays of the Day — a handful of calls per slate where the model has historically earned its opinion. The full methodology, and the live track record updated as games settle, is on the how it works page.

The data we work from

Every projection is built from publicly licensed sports data. The edge is in how the inputs are weighted and combined — not in secret data feeds. Here's the full source list, the same one that appears in the footer of every page:

None of this pipeline involves visitor data — it's schedules, stat rates, and sportsbook lines. For how we handle your data, see the privacy policy.

What graded in public actually means

The rule is simple: every game's score locks at first pitch. Once the first batter steps in, the projection and the line that were live go into the ledger unchanged. Nothing can be quietly nudged after the fact to look smarter. The entire archive is public — roll back to any past date on any board and see exactly what we said and what happened, in that order.

The same ledger that grades every call is the one that tunes the model. Each week the weights re-tune against what actually landed on a held-out test split — so a cold stretch hits the model directly, and it corrects. When we're on a bad run, you'll see it in the record before we say anything about it.

The honesty rules

These are the constraints we actually run the site by. They're not marketing copy; they're what the site does:

• Every call gets graded. No cherry-picking. A Play of the Day that loses stays in the record and counts against the hit rate, the same as one that wins.
• Efficient markets get labeled as such. We've backtested every market we cover. The moneyline, game totals, hits, home runs, RBI, runs, total bases — the books price them well enough that a sharp model can't reliably find an edge. We publish those as transparency boards, not plays. Two honest edges beat fifty manufactured ones.
• No locks. A single MLB game is close to a coin flip. A single prop isn't much better. Anyone promising a guaranteed winner is either fooling you or fooling themselves. We don't use that word.
• Misses get shown, not buried. When the model has a cold stretch, the record shows it. That's not a liability — it's what a real ledger looks like, and it's what tunes the model to do better.
• Nothing here is betting advice. Everything on this site is a research signal from a statistical model, published with the graded record so you can decide how much weight to give it. The call to wager, and everything that follows, is yours.

What we cover

MLB is the primary sport — the graded sample is deepest there, the model runs daily from Opening Day through the World Series, and the methodology page covers the full approach. We also cover golf (DataGolf strokes-gained projections, DFS lineup tools) and UFC (fight cards, model outputs, historical results). The record is deepest in MLB; golf and UFC are live and growing.

A daily recap engine — a morning article covering what the models said yesterday, what landed, and what missed — is coming when the foundational content is in place. A graded recap is only worth reading if the methodology it's built on is already trustworthy. That's what this phase of the build is about.

See it on today's board

Everything is free and the whole thing is graded in the open. Check the day's Plays of the Day, dig into the strikeout board where the model's edge is sharpest, or read exactly how the model works and how it's doing this season. Then form your own opinion about whether the numbers are worth trusting — which is the only thing we'd ever ask you to do.

Questions or data errors? Find us on X at @MatchWizAI. We correct in public.