Tuesday, September 29, 2026 · 16 picks · 16 games on the board · 0 priced
NFL / Picks
Week 4 Picks
Graded publicly
Public record
Picks started this week. 16 logged, 0 graded.
Grading begins after the first final. Every pick locks at kickoff and is
graded against the line it locked at.
Thursday, Oct 13pm to 6pm PT
Matchup · projected QBsWinner · win %Spread leanTier
PIT @ CLE
5:15 PM PT · O/U 38.5 (market)
PIT Aaron Rodgers · CLE Deshaun Watson
CLE +2.5
51.6% cover · 2.1 pt gap
LEAN
Sunday, Oct 4Before 12pm PT
Matchup · projected QBsWinner · win %Spread leanTier
TEN @ BAL
10:00 AM PT · O/U 43.5 (market)
TEN Cam Ward · BAL Lamar Jackson
TEN +11.5
52.6% cover · 3.4 pt gap
PLAY
NE @ BUF
10:00 AM PT · O/U 48.5 (market)
NE Drake Maye · BUF Josh Allen
NE +7
54.0% cover · 5.2 pt gap
PLAY
NYJ @ CHI
10:00 AM PT · O/U 42.5 (market)
NYJ Geno Smith · CHI Case Keenum not usual starter (usual: Caleb Williams)
CHI -3.5
52.5% cover · 3.3 pt gap
PLAY
IND @ WAS
6:30 AM PT · O/U 47.5 (market) · neutral site
IND Daniel Jones · WAS Jayden Daniels
IND -3.5
50.1% cover · 0.2 pt gap
LEAN
GB @ TB
10:00 AM PT · O/U 39.5 (market)
GB Jordan Love · TB Baker Mayfield
TB +4.5
54.3% cover · 5.7 pt gap
PLAY
DAL @ HOU
10:00 AM PT · O/U 47.5 (market)
DAL Dak Prescott · HOU C.J. Stroud
HOU -2.5
52.8% cover · 3.7 pt gap
PLAY
LA @ PHI
10:00 AM PT · O/U 45.5 (market)
LA Matthew Stafford · PHI Jalen Hurts
PHI +3
51.0% cover · 1.2 pt gap
LEAN
JAX @ CIN
10:00 AM PT · O/U 51.5 (market)
JAX Trevor Lawrence · CIN Joe Burrow
JAX +2.5
54.8% cover · 6.3 pt gap
PLAY
ARI @ NYG
10:00 AM PT · O/U 44.5 (market)
ARI Jacoby Brissett · NYG Jameis Winston not usual starter (usual: Jaxson Dart)
NYG +1.5
52.4% cover · 3.1 pt gap
PLAY
Sunday, Oct 412pm to 3pm PT
Matchup · projected QBsWinner · win %Spread leanTier
MIA @ MIN
1:05 PM PT · O/U 38.5 (market)
MIA Malik Willis · MIN Kyler Murray
MIA +11.5
53.7% cover · 4.9 pt gap
PLAY
LAC @ SEA
1:25 PM PT · O/U 42.5 (market)
LAC Justin Herbert · SEA Drew Lock
SEA -7
51.8% cover · 2.3 pt gap
LEAN
KC @ LV
1:25 PM PT · O/U 47.5 (market)
KC Patrick Mahomes · LV Kirk Cousins
LV +4.5
50.8% cover · 1.0 pt gap
LEAN
DEN @ SF
1:25 PM PT · O/U 46.5 (market)
DEN Bo Nix · SF Brock Purdy
DEN +3
50.2% cover · 0.3 pt gap
LEAN
Sunday, Oct 43pm to 6pm PT
Matchup · projected QBsWinner · win %Spread leanTier
DET @ CAR
5:20 PM PT · O/U 50.5 (market)
DET Jared Goff · CAR Bryce Young
CAR +3.5
51.9% cover · 2.5 pt gap
LEAN
Monday, Oct 53pm to 6pm PT
Matchup · projected QBsWinner · win %Spread leanTier
ATL @ NO
5:15 PM PT · O/U 48.5 (market)
ATL Michael Penix Jr. · NO Tyler Shough
ATL +3
51.0% cover · 1.3 pt gap
LEAN
How to read these picks.
Win % starts from the sportsbook spread and moves it 20% of the way
toward our ratings model, which accounts for the projected starting quarterback.
The spread becomes a probability through a curve fitted to past NFL games. Favorites
win about two of every three NFL games, and a 60% pick still loses four times in ten.
Spread leans are the side our model prefers against the current line.
Most sit between 50% and 54%, because the closing line is hard to beat, and we do not
inflate them. A lean needs better than 52.38% to be labelled PLAY: that is the break-even
at standard -110 odds. 8 of 16 spread leans clear the 52.38% break-even and are labelled PLAY.
The 20% weight is a starting assumption, not a measured edge. On 2023 data the model's
own weight measured close to zero, so 20% is generous. We refit it from the graded record
once 150 spread leans are final, and publish the result either way.
Totals are shown as the market number only. We do not pick overs or
unders. Updated Tue Sep 29, 7:01 AM PT.
Where the NFL data comes from.
Every rating and every graded pick uses the nflverse schedule file: closing
spreads, totals, final scores and starting quarterbacks for every game since 1999,
more than 7,300 of them. Before we used it, it had to pass a 10-point check. Final
scores land about 13.2 points from the closing spread, which is the
real scatter of NFL games, and the spread predicts the margin almost one for one.
We switched because our first source failed that check. In the old sportsbook
archive, the games where books disagreed most were the ones where the line predicted
the final score best:
| Book disagreement | Games | Scatter around the line |
| none | 116 | 11.80 |
| up to 1 point | 179 | 9.83 |
| up to 3 points | 331 | 8.40 |
| up to 8 points | 636 | 7.52 |
Real pregame lines cannot do that. In-game prices, posted after points were
scored, had leaked into the archive. Testing a model against them would have
flattered it. On the clean file, the test ran and gave a plain answer: our ratings
add little on top of the closing line. That is why the picks above lean on the
market and why most spread leans sit near 50%.
Also graded in public: the MLB first-inning board →
The Board
Sportsbook numbers · not our picks
MatchupBook lineBook O/UBook win %
ATL @ GB
FINAL ATL 35 - 14 GB
-
-
LAC @ BUF
FINAL LAC 16 - 24 BUF
-
-
CAR @ CLE
FINAL CAR 18 - 21 CLE
-
-
NYJ @ DET
FINAL NYJ 24 - 31 DET
-
-
HOU @ IND
FINAL HOU 17 - 19 IND
-
-
KC @ MIA
FINAL KC 24 - 10 MIA
-
-
TEN @ NYG
FINAL TEN 7 - 12 NYG
-
-
CIN @ PIT
FINAL CIN 27 - 30 PIT
-
-
SEA @ WSH
FINAL SEA 31 - 33 WSH
-
-
NE @ JAX
FINAL NE 6 - 35 JAX
-
-
ARI @ SF
FINAL ARI 30 - 36 SF
-
-
MIN @ TB
FINAL MIN 23 - 16 TB
-
-
BAL @ DAL
FINAL BAL 34 - 31 DAL
-
-
LV @ NO
FINAL LV 35 - 27 NO
-
-
LAR @ DEN
FINAL LAR 26 - 30 DEN
-
-
PHI @ CHI
FINAL PHI 7 - 27 CHI
-
-
Win % is the book's own spread converted to a probability, not our forecast.
A team laying 7 points wins about 70% of the time. Capped at 99%, because nothing
in football is certain.