2023 NFL season · final · regular season
Expected wins, 2023: what each team’s play was worth
The Bengals won 9 games; replayed with the same day’s form, their play was worth 6.2, the biggest gap in the league, and points scored and allowed put them at 8.0. The Commanders are the other way round: 4 wins from play worth 7.1.
| Team | Record | Replay wins | Pythagorean | Wins − replay |
|---|---|---|---|---|
| 13–4 | 13.8 | 13.3 | −0.8 | |
| 12–5 | 13.7 | 13.0 | −1.7 | |
| 12–5 | 11.0 | 12.9 | +1.0 | |
| 12–5 | 10.5 | 10.0 | +1.5 | |
| 11–6 | 13.6 | 12.0 | −2.6 | |
| 11–6 | 12.7 | 10.8 | −1.7 | |
| 11–6 | 12.3 | 10.8 | −1.3 | |
| 11–6 | 10.3 | 8.6 | +0.7 | |
| 11–6 | 8.2 | 9.4 | +2.8 | |
| 10–7 | 10.0 | 9.2 | 0.0 | |
| 10–7 | 8.8 | 9.2 | +1.2 | |
| 10–7 | 7.4 | 7.9 | +2.6 | |
| 9–8 | 9.7 | 8.7 | −0.7 | |
| 9–8 | 9.2 | 10.5 | −0.2 | |
| 9–8 | 8.6 | 9.4 | +0.4 | |
| 9–8 | 8.4 | 8.0 | +0.6 | |
| 9–8 | 7.8 | 9.2 | +1.2 | |
| 9–8 | 7.4 | 7.5 | +1.6 | |
| 9–8 | 6.2 | 8.0 | +2.8 | |
| 8–9 | 8.2 | 7.0 | −0.2 | |
| 8–9 | 7.9 | 8.5 | +0.1 | |
| 7–10 | 9.5 | 8.0 | −2.5 | |
| 7–10 | 9.4 | 7.0 | −2.4 | |
| 7–10 | 7.3 | 8.0 | −0.3 | |
| 7–10 | 5.3 | 5.8 | +1.7 | |
| 6–11 | 5.6 | 6.7 | +0.4 | |
| 6–11 | 3.5 | 4.5 | +2.5 | |
| 5–12 | 6.8 | 7.1 | −1.8 | |
| 4–13 | 7.1 | 4.3 | −3.1 | |
| 4–13 | 4.5 | 4.4 | −0.5 | |
| 4–13 | 4.5 | 5.4 | −0.5 | |
| 2–15 | 2.8 | 3.5 | −0.8 |
- Replay wins
- Summed over the season: how often the team wins each game when it’s replayed with the way both teams actually played.
- Coin-flip wins
- The same with only kicks, bounces, drops, conversions and calls re-rolled. The number that best predicts the rest of a season: correlation 0.437 against 0.378 for actual wins.
- Pythagorean
- From points scored and allowed: games × PF2.37 / (PF2.37 + PA2.37).
- Betting line
- Summed win chances from each game’s closing spread: what the market expected before kickoff.
Luck-adjusted standings
Each division: record on the left, replay wins on the right. A line that falls from left to right belongs to a team whose record ran ahead of its play.
MIA 11–6
BUF 11–6
NYJ 7–10
NE 4–13 12.313.65.34.5 RecordReplay wins
BAL 13–4
CLE 11–6
PIT 10–7
CIN 9–8 13.88.27.46.2 RecordReplay wins
HOU 10–7
IND 9–8
JAX 9–8
TEN 6–11 8.88.49.75.6 RecordReplay wins
KC 11–6
LV 8–9
DEN 8–9
LAC 5–12 12.77.98.26.8 RecordReplay wins
DAL 12–5
PHI 11–6
NYG 6–11
WAS 4–13 11.010.33.57.1 RecordReplay wins
DET 12–5
GB 9–8
CHI 7–10
MIN 7–10 10.58.67.39.5 RecordReplay wins
TB 9–8
NO 9–8
ATL 7–10
CAR 2–15 7.89.29.42.8 RecordReplay wins
SF 12–5
LA 10–7
SEA 9–8
ARI 4–13 13.710.07.44.5 RecordReplay wins
Adjust the assumptions
Change how much of each kind of play counts as luck, and every team’s coin-flip wins are recomputed for the whole season.
| # | Team | Wins | Coin-flip wins | Luck |
|---|---|---|---|---|
| 1 | 13 | 14.7 | −1.7 | |
| 2 | 11 | 13.8 | −2.8 | |
| 3 | 12 | 13.6 | −1.6 | |
| 4 | 12 | 11.6 | +0.4 | |
| 5 | 12 | 11.6 | +0.4 | |
| 6 | 11 | 11.5 | −0.5 | |
| 7 | 11 | 10.6 | +0.4 | |
| 8 | 11 | 9.7 | +1.3 | |
| 9 | 9 | 9.6 | −0.6 | |
| 10 | 10 | 9.0 | +1.0 | |
| 11 | 9 | 8.9 | +0.1 | |
| 12 | 7 | 8.8 | −1.8 | |
| 13 | 10 | 8.7 | +1.3 | |
| 14 | 11 | 8.5 | +2.5 | |
| 15 | 9 | 8.1 | +0.9 | |
| 16 | 9 | 7.9 | +1.1 | |
| 17 | 7 | 7.7 | −0.7 | |
| 18 | 9 | 7.6 | +1.4 | |
| 19 | 6 | 7.6 | −1.6 | |
| 20 | 9 | 7.6 | +1.4 | |
| 21 | 7 | 7.5 | −0.5 | |
| 22 | 9 | 7.4 | +1.6 | |
| 23 | 5 | 7.4 | −2.4 | |
| 24 | 7 | 7.3 | −0.3 | |
| 25 | 8 | 7.1 | +0.9 | |
| 26 | 8 | 7.0 | +1.0 | |
| 27 | 10 | 6.8 | +3.2 | |
| 28 | 6 | 6.7 | −0.7 | |
| 29 | 4 | 5.7 | −1.7 | |
| 30 | 4 | 4.3 | −0.3 | |
| 31 | 4 | 4.2 | −0.2 | |
| 32 | 2 | 3.4 | −1.4 |
2023 luck rankings · One-score games · Referee calls · Methodology