What happens before a pick ships
Every night the model prices each game, compares it against the market, and publishes only the edges.
Ingest — every game on the board
Before each slate the model pulls what moves a result: who is playing, how they have been playing, the conditions they are playing in, and how the market is pricing it across multiple sportsbooks.
Price — a probability for every market
Each matchup is priced by a machine-learning model trained on more than a decade of completed games and validated only on games it has never seen. The output is not a call on who wins; it is a probability for one specific market — this side covers, this total goes over — and from that, our own fair price for that line on the board. The model is trained separately per league and per market, and retrained weekly. We do not publish how it is built.
Compare — find where the market is wrong
Our price is compared against the market price, with the book's margin left in — the harder comparison, not the flattering one. Agreement means no bet; the market is right most of the time. A pick only exists when the gap clears the minimum threshold set for that market and the data behind it is fresh: a prediction built on stale inputs is dropped rather than staked.
Publish — timestamped, then graded
Every published pick goes out with its confidence, its edge and its unit size, and is timestamped on the public ledger before the games start. After the final it is graded — wins, losses, pushes, everything.
What the model gets wrong
Late news moves lines after we publish, thin-data spots are noisier, and variance is real — a winning model still loses plenty of individual bets, sometimes several in a row. We publish the losses anyway.
The model prices a game before the slate and does not read the news wire. A star scratched hours later can invalidate the math, and nothing downstream re-prices it. If the number has moved a long way from the price we published, treat the edge that justified the pick as gone.
Season openers, rookie debuts, post-trade rosters — when the history is thin the model is extrapolating, and it is worst exactly where it looks most confident. Predictions built on stale inputs are dropped rather than staked, which is one reason some nights the card is short.
A winning model still loses plenty of individual bets, sometimes several in a row. We’ve had losing months, and every one of them is on the public ledger. If a cold week would break your bankroll, bet smaller. The math only works over volume.
The skeptic's questions
If the model works, why sell it instead of just betting it?
Sportsbooks limit winning accounts quickly, which caps how much any model can earn betting alone — the better it does, the faster it gets cut off. Subscriptions do not have that ceiling. Both point the same direction: the record is the product, so the model has to keep winning in public.
Won't the edge disappear if thousands of people bet the same picks?
Lines do move when our subscribers bet. Every pick is published with the price we got, so you can see how far the number has travelled by the time you reach it — and if it has moved through the edge, the correct play is no play. The edge comes from a repeatable process, not from any single line staying soft.
Couldn't you fake the track record?
That's exactly why every pick is timestamped publicly before the game starts. A record you publish after results come in is marketing; a record timestamped before tip-off is evidence. Check the ledger yourself — the losses are right there.
Is this legal?
Yes. We're a sports analytics publisher — we don't take bets, hold your money, or operate as a sportsbook. Whether and where you bet is up to you and your local laws; we're for users 21+ in places where sports betting is legal.
Watch the pipeline work.
The model's highest-edge play lands in your inbox at 9 AM tomorrow. Free, every day.