Updated Sep 7, 2026 · Week 1 board.
Why the cross-check is dead this week
Screen one is the prop board. 212 player markets, every one of them Underdog. Screen two is that same board with a PrizePicks price beside it, and screen two is blank. Every comparison reads a gap of zero, because there is no second number in the room. A cross-check needs two books pricing the same guy. One moves, the other doesn't, and the space between them tells you something happened. An injury. A snap count. Rain. With one book that signal can't fire at all. We do have PrizePicks numbers from June sitting around. They're old enough that the only thing I use them for is matching a player to his team.
What's actually locked into the two late kickoffs
Cowboys at Giants closes Sunday at 8:20 PM Eastern. Broncos at Chiefs closes the entire week Monday at 8:15 PM Eastern. Those are the last two games on the board, so every late-swap decision you make runs through one of them. Here is what Underdog has on the main skill guys in both:
| Player | Team | Market | Underdog line |
|---|---|---|---|
| Dak Prescott | DAL | Pass yards | 260.5 |
| CeeDee Lamb | DAL | Receiving yards | 76.5 |
| Jaxson Dart | NYG | Pass yards | 213.5 |
| Cam Skattebo | NYG | Rush+rec yards | 70.5 |
| Patrick Mahomes | KC | Pass yards | 229.5 |
| Travis Kelce | KC | Receiving yards | 41.5 |
| Bo Nix | DEN | Pass yards | 227.5 |
| Courtland Sutton | DEN | Receiving yards | 42.5 |
Every number in that table is one book's line, as of today. Not a consensus. There is nothing on the other side of it to build a consensus with.
The numbers you can actually cross-check
Now put the props down. Both late games do have three books talking to each other on the spread and on the total. Cowboys-Giants runs anywhere from 2.5 to 3 points on the Giants depending where you look. The total sits between 47.5 and 48.5. Our own model likes the Giants better than the market does. We have them inside 1.47 points against a 2.5-point line. Call that an edge of 3.97 points at 83% confidence, meaning the odds our model gives itself of being right. On the total we project 53 against that 48.5 line. That is a 4.5-point edge at 73.7% confidence.
Broncos-Chiefs is flatter. All three books have the Chiefs laying 3, no daylight anywhere. The total sits between 43 and 43.5. Our model leans under 42.5, projecting 40.97, a small edge at 60.1% confidence. On the spread we project Denver within 0.56 points against the 3 points they are getting, which is a lean toward the Chiefs at 75.1% confidence.
Why a zero gap is not a zero-risk gap
The prop screen reads quiet. The screen I don't have might be screaming. That is the risk in this whole read. A gap of zero across 212 props doesn't mean the market is calm. It means I am watching one screen and calling it the room. Read missing information as no news and you will pay for it in the hour before kickoff.
What to do at your deadline
With props coming from one place, the game numbers above are the only part of this board carrying real cross-book signal. The under drift on Broncos-Chiefs argues against chasing a ceiling play late in that game. The model support on the Giants, on both markets, argues for treating that one as tighter than the spread makes it look. Neither is a record yet. Both are same-day leans, and I'd size them that way when you decide who to bench.
Live lines are on the odds page. Player markets are at player props. Build or fix a roster at the DFS hub. The general framework is in NFL DFS late-swap strategy and late-swap recovery strategy.
Both Screens Say
212 props, one book, zero second opinions, so that screen stays dark tonight. The one I'd actually keep open is the game board, where the Giants are pulling real model support and the Broncos-Chiefs total is leaning under. Here is my lean, and it is a lean. Fade the chalk ceiling play late in the Monday game, and don't panic about the Giants covering close. What flips it is PrizePicks posting a number on any of these players before either game kicks, because one second price turns 212 blind props into an actual cross-check.
