Whet for the NBA.
Not live yet — and we would rather tell you that than describe a model you cannot check.
Off-season. Nothing to show yet.
Whet has never issued a graded NBA pick. The scanner is not currently fetching NBA odds, because the season has not started and paying for odds nobody can bet is just a bill.
So there is no NBA record on this page, because there is no NBA record. The empty state below is read live from the same public ledger every other sport is measured against — it is not a placeholder we are choosing to show you.
No graded record
Scanned and graded before it is shown.
When NBA is switched back on, it goes through the same pipeline as everything else: the scanner prices every market it can find across books, and the settle job grades every issued pick against the real result the next morning.
It does not go on the live board at that point. It goes on issue-only — the picks are made and graded where you can check them, but they are not put in front of you as recommendations until the record justifies it.
The prices themselves get the same gates every sport gets: the two-way consensus rule — the fair line is built only from books quoting both sides of the market — a minimum of three independent books backing the best price, fitted on 21,782 graded MLB picks, rejection of any quote more than 15 cents off consensus, a 3% minimum edge on player props and 2% on mainlines, and an injury gate that removes props for players ruled out.
Steals and blocks have to clear a higher bar.
Per-game variance on steals, blocks and turnovers is far higher than on points, so a thin edge on one of them does not survive it. Whet rejects those three markets below 8% EV rather than the 3% it accepts elsewhere.
That rule is written and sits on the live scan path. It is dormant for the same reason everything else here is: nothing is being scanned to apply it to. It is on this page because it is real, and because it is the only sport-specific rule in Whet that is actually connected to anything.
No pace model, no back-to-back model.
This page used to describe a back-to-back filter dropping high-usage players around 10%, a pace adjustment blending both teams' possessions weighted toward the last ten games, a teammate-injury model reassigning usage to the player who historically absorbs it, and a blowout filter keyed to the Vegas spread.
The thresholds exist in the code. The data they read has never been connected to the scanner, so all of them take their defaults and change nothing. They have been removed from this page rather than left up as a description of the product.
If we build them properly, they will show up here with a graded record attached, the same as everything else on this site.
See what is live today?
Baseball and soccer are on the board now, with every graded pick — wins and losses — on the public ledger.