Prop Streak Lab

Know who clears the line this week.

The model's chance for every NFL prop, priced against the market, with every pick graded in public.

📧 Free daily picks email →

⚡ Value spots

market gives it 30%+, model is 15+ points better — overs and unders both
What counts as a value spot — both rules have to be true
  1. 1The market gives it at least a 30% chance. The price you'd pay has to be 30¢ or more. This is the rule that keeps penny longshots off the list — a 6¢ ask against an 11% model chance looks like a huge edge in percentage terms and is still a bad bet.
  2. 2Our model puts it 15 to 18 points higher. 30¢ ask → needs 45–48% · 50¢ → 65–68% · 70¢ → 85–88%. Not more: on the record so far, the bigger the gap, the worse NFL Value picks have done (18–20 points up went 8–15 in weeks 5–6) — a gap that big has more often been the model missing a role change the market already knew. And if any line of a prop is more than 20 points up, that whole prop is left off. The gap between the two, in percentage points, is the edge shown in the Best price column.
Both rules are tested against the best price in the row — the cheaper of the live Polymarket US price and Kalshi's, the price you'd actually pay — so the edge shown is always the edge it qualified on, and a prop drops off the list as soon as the price moves against it. A price only counts when at least $25 is offered at it (or within 2¢) — in a near-empty market your bet would move the price — so a thin price is skipped for the other exchange's, or the prop is left off. One line per player-prop (the rung with the biggest edge). Both rules are checked on the over and the under of every market separately, so either side can qualify, and a prop also needs 6+ effective games of history behind it. Everything clearing both rules is listed — no fixed top 25 — ranked by that edge, and the list re-checks itself about once a minute. Props waiting on injury news stay off this list and Top 25 Surest: when the player, or a teammate whose snaps decide his role (a quarterback for the whole passing game), is questionable or doubtful — until ESPN shows him active or out (on game day, the inactives 90 minutes before kickoff). Ruled out: his teammates' props come back, with the model's numbers.

The board scan fills this in. Needs the hosted website (the Polymarket API can't be reached from a local file).

🔒 Top 25 Surest

Scans every player prop on this week's Polymarket board and shows the 25 highest model chances you can bet (a live price with $25+ offered near it), each next to its best price on Polymarket US or Kalshi — the surest plays by the numbers, regardless of payout. Especially handy if you're holding credits you can't withdraw without trading: these are the props most likely to resolve as wins, so clearing that requirement lets you cash out your balance. Each pick is recorded before kickoff and graded in the Past picks tab. Needs the hosted website (the Polymarket API can't be reached from a local file).