4th down and 2-point decisions
Every 4th down and every try after a touchdown this season, graded against Ben Baldwin's nfl4th model (MIT license), reproduced exactly. It is nfl4th's recommendation, not this site's model, and it is not adjusted for the teams playing. Try any situation in the calculator.
nfl4th leans toward going for it. It recommends going on 41.8% of this season's 1,206 evaluated 4th downs; coaches went on 16.3%.
So most flagged decisions are kicks the model would have gone for: 105 of 112 (93.8%). They are also the plays where the model is least certain: 87.5% had nfl4th's top two options within 5 points of win probability.
Win probability lost to mistakes
| Rk | Coach | Team | Decisions | Mistakes | Clear mistakes | WP lost | Per decision |
|---|---|---|---|---|---|---|---|
| 1 | Jesse Minter | BAL | 54 (38 + 16) | 7 | 3 | 35.4 | 0.66 |
| 2 | Sean Payton | DEN | 55 (43 + 12) | 7 | 2 | 28.7 | 0.52 |
| 3 | Shane Steichen | IND | 53 (41 + 12) | 6 | 1 | 22.5 | 0.42 |
| 4 | Liam Coen | JAX | 45 (31 + 14) | 6 | 1 | 21.9 | 0.49 |
| 5 | Zac Taylor | CIN | 41 (28 + 13) | 6 | 1 | 18.6 | 0.45 |
| 6 | Mike Macdonald | SEA | 57 (43 + 14) | 6 | 0 | 17.0 | 0.30 |
| 7 | Jim Harbaugh | LAC | 50 (41 + 9) | 4 | 1 | 15.2 | 0.30 |
| 8 | Jonathan Gannon | ARI | 53 (43 + 10) | 3 | 2 | 14.8 | 0.28 |
| 9 | Dan Campbell | DET | 47 (33 + 14) | 2 | 1 | 14.2 | 0.30 |
| 10 | Brian Schottenheimer | DAL | 47 (31 + 16) | 4 | 0 | 13.8 | 0.29 |
| 11 | Robert Saleh | TEN | 41 (35 + 6) | 5 | 0 | 13.7 | 0.33 |
| 12 | Kellen Moore | NO | 56 (40 + 16) | 5 | 0 | 12.5 | 0.22 |
| 13 | DeMeco Ryans | HOU | 54 (40 + 14) | 3 | 1 | 11.6 | 0.21 |
| 14 | Sean McVay | LAR | 49 (34 + 15) | 5 | 0 | 11.4 | 0.23 |
| 15 | Kyle Shanahan | SF | 48 (27 + 21) | 4 | 0 | 10.9 | 0.23 |
| 16 | Matt LaFleur | GB | 52 (39 + 13) | 4 | 0 | 10.8 | 0.21 |
| 17 | John Harbaugh | NYG | 51 (39 + 12) | 4 | 0 | 10.4 | 0.20 |
| 18 | Dave Canales | CAR | 49 (35 + 14) | 4 | 0 | 9.9 | 0.20 |
| 18 | Mike McCarthy | PIT | 58 (45 + 13) | 4 | 0 | 9.9 | 0.17 |
| 20 | Ben Johnson | CHI | 60 (43 + 17) | 4 | 0 | 9.7 | 0.16 |
| 21 | Aaron Glenn | NYJ | 53 (42 + 11) | 3 | 0 | 9.0 | 0.17 |
| 22 | Andy Reid | KC | 42 (28 + 14) | 3 | 0 | 8.8 | 0.21 |
| 23 | Mike Vrabel | NE | 54 (37 + 17) | 3 | 0 | 8.6 | 0.16 |
| 24 | Kevin O'Connell | MIN | 50 (42 + 8) | 2 | 0 | 7.7 | 0.15 |
| 25 | Dan Quinn | WAS | 60 (45 + 15) | 2 | 0 | 7.6 | 0.13 |
| 26 | Klint Kubliak | LV | 55 (39 + 16) | 1 | 1 | 7.1 | 0.13 |
| 27 | Nick Sirianni | PHI | 56 (44 + 12) | 1 | 0 | 4.7 | 0.08 |
| 28 | Todd Bowles | TB | 57 (46 + 11) | 2 | 0 | 4.1 | 0.07 |
| 29 | Todd Monken | CLE | 56 (47 + 9) | 1 | 0 | 2.2 | 0.04 |
| 30 | Sean McDermott | BUF | 49 (27 + 22) | 1 | 0 | 2.0 | 0.04 |
| 31 | Jeff Hafley | MIA | 32 (27 + 5) | 0 | 0 | 0.0 | 0.00 |
| 31 | Raheem Morris | ATL | 45 (33 + 12) | 0 | 0 | 0.0 | 0.00 |
Decisions counts every graded 4th down and try (in brackets, 4th downs plus tries): supported by nfl4th, not in a game already decided. A clear mistake is one where nfl4th's top two options were 5 or more points of win probability apart, so the model was not torn. Select a column heading to sort by it. WP lost adds up the mistakes, in points of win probability; per decision divides it by all decisions, so a coach with few decisions shows it. Equal totals share a rank. Each game counts for the head coach who coached it, so an interim coach is ranked for his own games.