NFL Prop Intelligence: timestamped prop-market movement, last-10 hit rates and injury context — packaged as a file you drop into your own AI, with the questions to ask it. Not picks. The data, and the method.
Week 1 pack publishes Sunday morning · 21+ · Informational only
One file per slate. Every curated prop with its opening line, its latest line, every move in between, and the last-10 history behind it.
ChatGPT, Claude, or any assistant with file upload and data analysis. The file is structured so a plain question gets a precise answer.
The prompt card gets you a full read in five minutes. After that, it's your analyst: trends by team, by market, by kickoff window.
Real rows from the Week 1 pack, captured from the public prop market multiple times a day since the lines opened.
| Player | Market | Open | Latest | Δ line | Snaps | L10 | Last 5 | Context |
|---|---|---|---|---|---|---|---|---|
| Michael Mayer | Rec Yds | 17.5 | 39.5 | +22.0 | 15 | 0.30 | U U U O O | Line up 22 since open; teammate TE Brock Bowers absent from board since 09/09 — report: Out (knee) — likely depth-chart driven |
| Woody Marks | Rush Yds | 30.5 | 24.5 | −6.0 | 22 | 1.00 | O O O O O | Line down 6 since open; 10 of last 10 over current line |
| George Pickens | Receptions | 5.5 (+124) | 4.5 (−158) | −1.0 | 21 | 0.70 | O U O O O | Line down 1; Over price shortened 82c; 7 of last 10 over current line |
| Dawson Knox | Rec Yds | 14.5 | 13.5 | −1.0 | 18 | 0.90 | O O U O O | Line down 1 since open; 9 of last 10 over current line |
| Josh Downs | Receptions | 3.5 (−158) | 4.5 (+124) | +1.0 | 21 | 0.20 | U U O U U | Line up 1; Over price lengthened 82c; 2 of last 10 over current line |
Δ line = latest minus opening. Snaps = number of market snapshots behind the row. L10 = share of the player's last 10 games clearing the current line. Every column is defined in the data dictionary that ships with the pack.
Opening line, latest line, the price on both sides, and the exact time of the last move — so you can tell a Tuesday opener from a Sunday-morning sharpen.
How often the player cleared the current line in his last ten games, with the last five spelled out. Verified against public box scores; blank when we can't verify.
Official report status plus our own tell — a player disappearing from the board between snapshots — and which teammates' lines moved after. The "why" behind the move.
Every pack ships with a prompt card: six questions that turn a spreadsheet into a conversation. Ask them in order and your AI walks you through the whole slate. Then ask your own.
Try before you buy: download the free 25-row sample and run the six questions on it.
No. Nothing in a pack recommends a wager. It reports what the market did, what the player did, and what changed on the injury report — stated facts with timestamps. What you do with it is yours.
Any AI service with file-upload and data-analysis capability. Availability varies by provider and plan; free tiers often can't read files. We test every pack in ChatGPT and Claude before it ships.
AtlasEdge Sports captures the public prop market on a fixed schedule — several snapshots a day during the week, every 30 minutes on Sunday morning. Historical results come from public box scores. We never publish a number we can't trace to a snapshot.
A full Sunday board is over 10,000 rows; no AI reads that well and neither do you. We keep the props with real movement, a strong last-10 record, or a status change behind them — around 200 rows that actually say something.
A preliminary Sunday pack publishes Saturday night; the final refreshes Sunday morning after the last capture, before the 1 PM window. Day packs publish the afternoon of the game.
Digital files can't be returned, but if a pack fails our own QA or a file won't open, tell us and we'll make it right.