New England Patriots at Buffalo Bills

Sunday Early · Sun, Oct 4, 1:00 PM ETNE @ BUF
New England Patriots

Away

New England Patriots

1-2

Buffalo Bills

Home

Buffalo Bills

3-0

Model Win Probability

New England Patriots · Away

31%

Model
separation
38%

Buffalo Bills · Home

69%

Signal: Moderate· components agree · 100% feature coverage
Elo component
27% NE
Efficiency component
34% NE
Feature coverage
100%
Data cutoff
2026-09-29T03:18:28.737160Z
Model
pt-nfl-ensemble-v0.3

Signal strength describes model agreement and data quality, not certainty or wagering value.

Team Efficiency Comparison

61
100
Offense
74
97
Passing
32
97
Rushing
87
45
Defense
81
90
Pass Rush
29
97
Situational

0–100 league percentile across all 32 teams. Defensive ratings are inverted so a higher number always means a better unit.

Blended Team Metrics

NEMetricBUF
0.040EPA / Play0.162
44.7%Success Rate48.9%
0.157Dropback EPA0.241
-0.128Rush EPA0.029
13.1%Explosive Play Rate15.7%
41.5%3rd Down Conversion51.3%
48.2%Red Zone TD Rate70.4%
-0.068Def. EPA / Play Allowed0.022

Prior-season form is blended with current-season play and its weight decays as the current sample grows. Metrics without enough data are shown as “—”, never as zero.

Game Context

Venue
Highmark Stadium
Kickoff
Sun, Oct 4, 1:00 PM ET
Roof
outdoors
Surface
a_turf
Away rest
7 days
Home rest
7 days
Site
Buffalo Bills home game

Player Impact

Drake Maye

QB

2026 Weeks 1, 2, 3

CMP/ATT
51/80
Pass yds
585
Pass TD
1
INT
6
Y/A
7.3
Sacks
9

Rhamondre Stevenson

RB

2026 Weeks 1, 2, 3

Carries
31
Rush yds
116
YPC
3.7
Rush TD
0
Targets
11
Rec
9
Rush share
36.5%

TreVeyon Henderson

RB

2026 Weeks 2, 3

Carries
24
Rush yds
99
YPC
4.1
Rush TD
1
Targets
1
Rec
1
Rush share
28.2%

Mack Hollins

WR

2026 Weeks 1, 2, 3

Targets
16
Rec
12
Rec yds
156
YPR
13
Rec TD
0
Target share
19.5%

DeMario Douglas

WR

2026 Weeks 1, 2, 3

Targets
14
Rec
7
Rec yds
67
YPR
9.6
Rec TD
0
Target share
17.1%

Scoring by Quarter

Sample · 3 games

Away · New England Patriots

Q1
7
Q2
13
Q3
6
Q4
10

Home · Buffalo Bills

Q1
17
Q2
44
Q3
10
Q4
30

Small sample — treat quarter splits as descriptive context, not a trend.

Latest News

Analytical Angles

Passing profile · NE

passing

New England Patriots ranks in the 74th league percentile in blended dropback EPA entering Week 4.

Defense profile · NE

efficiency

New England Patriots ranks in the 87th league percentile in blended defensive EPA allowed; higher percentile means less EPA conceded.

Passing profile · BUF

passing

Buffalo Bills ranks in the 97th league percentile in blended dropback EPA entering Week 4.

Defense profile · BUF

efficiency

Buffalo Bills ranks in the 45th league percentile in blended defensive EPA allowed; higher percentile means less EPA conceded.

Statistical context only. Picks Terminal publishes matchup analysis, not wagering advice.

Keys to the Game

  • Margin-aware Elo favors Buffalo Bills at 72.8%.
  • Rolling efficiency favors Buffalo Bills at 66.4%.
  • The calibrated estimate is 69%; it is a probability, not a certainty.

What Could Flip It

  • The statistical sample includes 3 current-season games for New England Patriots and 3 for Buffalo Bills; prior-season form is still blended in.
  • The components agree, but a single game still has substantial variance.
  • The first official Week 4 practice/game-status report was unavailable at this cutoff; documented quarterback news is listed separately and is not explicitly priced into this statistical model.

Model Analysis

New England Patriots at Buffalo Bills: the v0.3 ensemble places Buffalo Bills at 69% after a 70% margin-aware Elo and 30% rolling efficiency blend with strictly out-of-fold calibration. The statistical sample includes 3 current-season games for New England Patriots and 3 for Buffalo Bills; prior-season form is still blended in. No odds, moneylines, spreads, public picks or market consensus are model inputs.

Forecast

Model pick: Buffalo Bills (69%). Signal: Moderate; model agreement and coverage, not wagering value.

Sources & Methodology

Win probability: pt-nfl-ensemble-v0.3 — a deterministic ensemble of a margin-aware Elo component and a regularized efficiency component built from pregame team differentials (EPA per play, dropback and rush EPA, success rate, explosive rate, sack performance, home field and rest). The blended probability is calibrated on training seasons only and validated by walk-forward backtest against completed results. All inputs come from public play-by-play and schedule data. No external market data of any kind is used anywhere on this page.