I don't know the right apportionment of credit/blame between players and coaches in a close loss. And while I understand that every minute of the game should count equally in theory, performance down the stretch had an outsize impact for this team. So .625% or whatever ain't it.
The table below is populated with data from KenPom. Against every opponent with a pulse (Georgia might be a stretch), it has Northwestern's win probability entering the game, win probability with 5 minutes to go, and the outcome. I'm sure there are individual data points to quibble with, but keep in mind we've got a 25-game sample:
| Team | Loc | Starting Win Prob | 5-Min Win Prob | Win Prob Change | Outcome | Final Win Prob Change |
|---|
| Prov. | N | 59.5% | 8.7% | -50.8% | L | -8.7% |
| Georgia | N | 86.3% | 99.5% | 13.2% | W | 0.5% |
| Wake | A | 56.1% | 34.4% | -21.7% | L (OT) | -34.4% |
| MD | A | 41.9% | 70.0% | 28.1% | W | 30.0% |
| MSU | H | 48.5% | 31.5% | -17.0% | L | -31.5% |
| PSU | H | 77.8% | 68.8% | -9.0% | L | -68.8% |
| OSU | A | 22.8% | 1.9% | -20.9% | L | -1.9% |
| MD | H | 73.2% | 58.9% | -14.3% | L (2OT) | -58.9% |
| MSU | A | 18.3% | 50.2% | 31.9% | W | 49.8% |
| Wisky | H | 51.4% | 12.6% | -38.8% | L | -12.6% |
| Purdue | A | 9.7% | 1.3% | -8.4% | L | -1.3% |
| Mich | A | 22.4% | 78.5% | 56.1% | L | -78.5% |
| Illinois | H | 37.1% | 63.3% | 26.2% | L | -63.3% |
| Rutgers | H | 77.9% | 98.4% | 20.5% | W (OT) | 1.6% |
| Nebby | A | 70.5% | 100.0% | 29.5% | W | 0.0% |
| Indiana | H | 52.7% | 89.1% | 36.4% | W | 10.9% |
| Illinois | A | 17.1% | 16.5% | -0.6% | L | -16.5% |
| Purdue | H | 34.5% | 1.1% | -33.4% | L | -1.1% |
| Minny | A | 57.3% | 0.9% | -56.4% | L | -0.9% |
| Nebby | H | 90.0% | 99.8% | 9.8% | W | 0.2% |
| PSU | A | 40.3% | 40.5% | 0.2% | L | -40.5% |
| Iowa | A | 14.8% | 0.0% | -14.8% | L | 0.0% |
| Minny | H | 75.4% | 98.4% | 23.0% | W | 1.6% |
| Nebby | N | 75.5% | 51.7% | -23.8% | W | 48.3% |
| Iowa | N | 17.4% | 0.0% | -17.4% | L | 0.0% |
| | | | | | |
| Total | W | 12.3 | 11.8 | -0.5 | 9 | -2.8 |
| Total | L | 12.7 | 13.2 | 0.5 | 16 | 2.8 |
That's a massive amount of variation from expected record due to how the team played at the end of games, even if it's not that surprising. For comparison, Penn St's expected record in similar games was 9.4-15.6, its expected record with 5 minutes left was 8.5-16.5, and its final record was 9-16 -- nothing too extreme in that profile. (That was the first team I checked because they played a lot of the same teams and had almost identical advanced metrics. I'm happy to figure this out for others, though.) As bad as this is, it doesn't really suggest that we should blame Collins for 6-8 extra losses, either.