Buffalo Bills at New York Jets. No pick on the spread and no pick on the total.
The picks come first on this page. The market's number, our number and the gap between them sit under them as MODEL CONTEXT: that is the input our analyst panel argued from, not a call of its own. Tiers are how decisively the panel agreed, never a win probability.
Who is on the injury report
- J. Hancock CB Questionable, quad Sep 10
- T. Johnson RB Questionable, hamstring Sep 10
- T. Sanders DL Questionable, knee Sep 10
- J. Ossai DL Out, foot Sep 10
- K. Nwangwu RB Out, back Sep 11
- T. Patrick WR Out, groin Sep 10
- D. Ponds CB Doubtful, calf Sep 10
- B. Hall RB Injured, groin Sep 10
- I. Davis RB Injured, knee Sep 10
- K. Sadiq TE Injured, groin Sep 10
Latest daily scan (2026-09-13), real report rows only, as the source lists them; the report does not say who starts. Injuries are shown here as context and are not an input to the number on this page: the game number is the rating difference plus home field. The model's quarterback continuity term (about 2.8 points when the expected starter does not start) lives in the season expected wins walk, not in this read.
Model context
The numbers the panel argued from. Our models price a game, the panel makes the call, and the picks above are the call. Nothing in this block is a pick. Where the panel's written case cites a raw number of its own, the calibrated number is the one printed here.
Context, not a call. The spread number is the EL NFL Rating v3 difference plus 2.122 points of home field. The totals number is a registered candidate that loses to the closing total: across 1,087 held out games it missed the combined score by 10.46 points on average and the closing total missed by 10.189, ahead of us in all 4 of those seasons, so it never posts to the board and never enters the record on its own.
Over and under splits
Descriptive context on this game, every line with the games it counts. Filters like these did not hold up as predictors in our testing, so none of them is a reason a pick wins and none of them enters a tier.
- the total went over in 4 of their last 5 (n=5, 2025)
- the total went over in 5 of their last 10 (n=10, 2025)
- the total is 2-7 to the over in road night games (n=9, 2023 to 2025)
- the total is 10-4 to the over in turf divisional games (n=14, 2023 to 2025)
- the total went over in 4 of their last 5 (n=5, 2025)
- the total went over in 7 of their last 10 (n=10, 2025)
- the total is 8-4 to the over as a favorite after a straight up loss (n=12, 2023 to 2025)
- the total is 15-8 to the over on turf after a straight up loss (n=23, 2023 to 2025)
Trends
Descriptive context, every line with the games it counts. Filters like these did not hold up as predictors in our testing, so none of this is in the confidence read and none of it is a reason a pick wins. Situational splits look back 3 seasons and need at least 8 decided games to be shown at all.
- covered 2 of their last 5 (n=5, 2025)
- the total went over in 4 of their last 5 (n=5, 2025)
- 11-1 straight up outdoors after a straight up loss (n=12, 2023 to 2025)
- covered 0 of their last 5 (n=5, 2025)
- the total went over in 4 of their last 5 (n=5, 2025)
- lost 5 straight
- 1-10 straight up in underdog grass games (n=11, 2023 to 2025)
Every split we can compute on this game
- won 4 of their last 5 (n=5, 2025)
- covered 2 of their last 5 (n=5, 2025)
- the total went over in 4 of their last 5 (n=5, 2025)
- won 7 of their last 10 (n=10, 2025)
- covered 5 of their last 10 (n=10, 2025)
- the total went over in 5 of their last 10 (n=10, 2025)
- 11-1 straight up outdoors after a straight up loss (n=12, 2023 to 2025)
- 16-2 straight up in home day games (n=18, 2023 to 2025)
- 8-1 straight up in day games after a straight up loss (n=9, 2023 to 2025)
- 8-1 against the number in day games after a straight up loss (n=9, 2023 to 2025)
- 8-1 straight up in home divisional games (n=9, 2023 to 2025)
- 21-3 straight up in home turf games (n=24, 2023 to 2025)
- 7-1 straight up at home after a straight up loss (n=8, 2023 to 2025)
- 22-4 straight up at home (n=26, 2023 to 2025)
- 22-4 straight up in home outdoor games (n=26, 2023 to 2025)
- 20-4 straight up in home favored games (n=24, 2023 to 2025)
- 9-2 straight up as a favorite after a straight up loss (n=11, 2023 to 2025)
- 25-6 straight up in outdoor turf games (n=31, 2023 to 2025)
- 24-6 straight up in outdoor day games (n=30, 2023 to 2025)
- 20-5 straight up in day turf games (n=25, 2023 to 2025)
- 12-3 straight up after a straight up loss (n=15, 2023 to 2025)
- 23-6 straight up in favored day games (n=29, 2023 to 2025)
- 11-3 straight up in turf divisional games (n=14, 2023 to 2025)
- 29-8 straight up in favored outdoor games (n=37, 2023 to 2025)
- the total is 2-7 to the over in road night games (n=9, 2023 to 2025)
- won 0 of their last 5 (n=5, 2025)
- covered 0 of their last 5 (n=5, 2025)
- the total went over in 4 of their last 5 (n=5, 2025)
- won 3 of their last 10 (n=10, 2025)
- covered 4 of their last 10 (n=10, 2025)
- the total went over in 7 of their last 10 (n=10, 2025)
- lost 5 straight
- failed to cover 5 straight
- 1-10 straight up in underdog grass games (n=11, 2023 to 2025)
- 1-8 straight up in road divisional games (n=9, 2023 to 2025)
- 1-7 straight up in underdog night games (n=8, 2023 to 2025)
- 2-13 straight up in underdog divisional games (n=15, 2023 to 2025)
- 2-11 straight up in day divisional games (n=13, 2023 to 2025)
- 3-15 straight up in road underdog games (n=18, 2023 to 2025)
- 3-13 straight up on grass (n=16, 2023 to 2025)
- 3-12 straight up in road grass games (n=15, 2023 to 2025)
- 3-12 against the number in underdog divisional games (n=15, 2023 to 2025)
- 3-11 straight up in divisional games after a straight up loss (n=14, 2023 to 2025)
- 3-11 straight up in outdoor grass games (n=14, 2023 to 2025)
- 3-11 straight up in turf divisional games (n=14, 2023 to 2025)
- 3-11 against the number in turf divisional games (n=14, 2023 to 2025)
- 4-14 straight up in divisional games (n=18, 2023 to 2025)
- 4-14 straight up in outdoor divisional games (n=18, 2023 to 2025)
- 2-7 against the number in road divisional games (n=9, 2023 to 2025)
- 8-27 straight up as an underdog (n=35, 2023 to 2025)
The context we can hand you free
Rest. New York Jets 7 days · Buffalo Bills 6 days Rest comes straight out of the schedule table, so it costs nothing to give away. Be careful reading it as an edge: the rest differential layer was tested for the v3 model and REJECTED, because it made out of sample margin error worse rather than better. It is context, not a signal.
Venue. Roof and surface on file: outdoors · fieldturf. Neither roof nor surface is a v3 model input, and the model document says so under what it does not factor. They are here because you asked what the game looks like, not because we priced them.
How the pick is made, and what the tier means
The pick comes from the analyst panel. Three seats work the same evidence pack and the week's news independently, name a side on the spread and on the total, and the tier on each pick above is how decisively they agreed: high, medium, low, or no pick when they did not. A tier is never a win probability.
Our number, in MODEL CONTEXT above, is the EL NFL Rating v3 difference plus 2.122 points of home field. The home field figure is the 2.122 intercept of the EL NFL Rating v3 second-stage margin model (docs/el-nfl-rating-v3.md); the per-team home field variant was tested in that same document and rejected for making out of sample error worse, so one league wide constant is the documented choice. That number is an input the panel argues from, and the panel can and does take the other side of it.
Said plainly, because the model's own document says it: v3 is a margin and expected wins engine that beat our prior model on margin error in four straight held out seasons, and it is not a proven spread beater. Treat this as research, not a guarantee.
Line movement
Not enough capture history on this game yet. Movement lands here once the odds capture has run more than once against this number before kickoff.
The two numbers underneath
| Team | EL NFL Rating v3 | Side |
|---|---|---|
| Buffalo Bills | 3.62 | Away |
| New York Jets | -4.54 | Home |
The rating is points versus a league average opponent on a neutral field. The difference plus 2.122 points of home field is our number above. Full board: the NFL power ratings.
Provenance: schedule, rest days, roof, surface, and finals from edgelabs.nfl_games (loaded by scripts/el_nfl_games_load.py from nflverse schedules); the date is the schedule's gameday, a calendar date, so no kickoff time is shown that we did not capture. The line is the latest capture for this game from edgelabs.nfl_lines with the book named. The movement block reads our permanent capture log (edgelabs.nfl_lines_history), one book across both ends and never a capture taken at or after kickoff, so the second number is the last pregame line; a game the log has captured only once shows no movement rather than an invented one. Our number is the EL NFL Rating v3 (edgelabs.el_nfl_rating, season 2026, model_version v3, docs/el-nfl-rating-v3.md) difference plus 2.122 points of home field, which is the 2.122 intercept of the EL NFL Rating v3 second-stage margin model (docs/el-nfl-rating-v3.md), and it is context rather than the call. The ATS pick and the total pick come from the analyst panel (edgelabs.nfl_sme_picks, migration 0091), three seats per game synthesized into one call with a tier; a tier is how decisively the panel agreed and is never a win probability. The write-up is assembled from stored rows only: the case from the lock row, the series from edgelabs.nfl_games from 2013 forward, and the scoring profiles from completed games with the number of games each average carries. A section with no data behind it is left out instead of filled in, it refreshes while the game is pregame, and it can never be edited once the game has kicked off. Trends are descriptive only, computed by the same shared module the write-up uses: straight up, against the number and over or under, from edgelabs.nfl_games (which carries the closing spread, the total, the roof, the surface and the divisional flag back to 2002), situational splits over three seasons with at least eight decided games, every line carrying the games it counts. Filters like these did not hold up as predictors in our testing, so none of them enter the confidence read. The totals block is a read and not a pick: our number is EL NFL Totals v1 (edgelabs.el_nfl_totals, season 2026, model_version v1, docs/el-nfl-totals-v1.md), a registered candidate that loses to the closing total, so it never posts to the board or the record; the market total beside it is the same capture the spread comes from before kickoff and the frozen consensus the model priced against after it, and the gap is measured against whichever number is printed. Research and context, never a guarantee, 21+.