Craftypicks exists because the sports betting information business is mostly broken. Someone sells you a pick, hides the record, and blames you when it loses. This site sells nothing and picks nothing. It prices the board, publishes its own number beside the market's, and keeps score of how close that number was.
This is the honest version of what the system does. Craftypicks has no injury model, no simulation engine, and no opinion about who's going to win tonight. It has one thing: the price at every sportsbook, and the math to tell when one of them is out of line.
Each morning the scanner reads the moneyline, spread and total at every major US book for every game in season. A game priced by fewer than 6 books is skipped — a thin consensus isn't a consensus, it's noise.
A book's price includes its margin, so the two sides of a market always add up to more than 100%. Removing that margin proportionally turns each book's price into its honest estimate of the real odds.
Those honest estimates get averaged into a consensus fair price. The book we'd actually be betting is left out of its own average — include it and the outlier drags the "fair" price toward itself and manufactures an edge that isn't there.
Every game gets its number, the best price on it, and the gap between the two — including the ones where there is no gap, which is most of them. Then every published number is scored against the final result and the accuracy stays on the page.
What this method can and can't do. It can find spots where one book is slow, has taken lopsided action, or simply disagrees with the market — and those spots are genuinely profitable over a large enough sample. It cannot tell you a starter just got scratched, that a team is on a brutal travel spot, or that the total is wrong because of the wind. If you have that information, it belongs on top of what you see here, not underneath it.
What we don't do: sell picks, rank "locks," tell you how much to bet, or quietly restate a number after the result is in. Nothing here is a recommendation. On most nights our number and the market's agree closely enough that there is nothing worth looking at, and the board says so instead of inventing something.
Every win probability and every projection this site publishes is stored the morning it goes up, then settled against the final result. Nothing is edited afterwards.
Each board carries its own score, because they are not the same problem: a strikeout projection is judged by how many strikeouts it missed by, a win probability by its Brier score against the market's.
| Board | Scored by | Compared against |
|---|---|---|
| Game boards | Brier score | The market's own price |
| Pitcher strikeouts | Average miss, in K | The posted line |
| Player props | Calibration bands | How often it actually hit |
Everything the system produces is on these pages. There isn't a paid group getting different or earlier numbers — that's the exact scam this was built against.
There is no list. Nothing here asks for an email, so there is nothing to sell, no "partner offers," and no third-party sportsbook spam.
A scanner, a static site, and a free host. There's no expensive infrastructure that needs a paywall to justify itself, so there's no pressure to build one.
No. Craftypicks is not a sportsbook. We don't take wagers, hold money, or process any transaction. The site publishes prices and how accurate they turned out to be — what you do with that is entirely your own business.
9:00 AM ET every day. The scanner runs, the boards get priced, and the site rebuilds itself — there's no human deciding what to include or leave out after seeing the results. The player prop boards refresh again in the afternoon, once lineups are posted.
Less than it was, and often nothing. The whole point of a gap is the distance between one book's number and the rest of the market; once the number moves, that distance is gone. Every board carries the time it was priced for exactly this reason.
To see the number shown, yes — a gap exists at that book's price, not at the market average. This is the unglamorous truth of price-based betting: line shopping across a few books is most of the work. We don't get paid by any book and don't link to them.
Because they're real. A market price is a very good forecast, and beating it by a hair on a few hundred games is what a working model actually looks like. Where our number is no better than the market's, the page says so rather than hiding it.
Reading the board faster than you could by opening six sportsbooks. Where every book sits, where the best number is, what our own model thinks, and how often that model has been right. Nothing on the page is telling you to do anything.
Yes. If that ever changed, the accuracy history would stay public and free regardless — paywalling the scorecard would defeat the entire point of keeping it.
Sports betting is entertainment with a negative expected value for almost everyone who does it. A better price reduces how much you lose — it does not turn betting into income, and nobody should be treating it that way.
If you're betting to make back losses, betting money set aside for rent or bills, hiding it from people close to you, or feeling anxious when you're not in action — those aren't strategy problems. Step away and talk to someone.
1-800-GAMBLER
National Problem Gambling Helpline — 24/7, all 50 states.
Text 800GAM to 800177
If you'd rather not talk on the phone.
Most sportsbooks offer deposit limits, time-outs, and self-exclusion in their account settings. Using them is not an admission of anything — it's the same thing as flat staking, just applied to your time instead of your money.
Every game priced, our number beside the market's, and the accuracy score on each board kept in public.