## Unbelieveable: A Win Probability Analysis of 2015's Craziest College Football Weekend

Week 9 of this college football season figured to deliver some juicy action, but no one could have predicted the onslaught of dramatic games that we got. At the end of the madness, most ranked teams avoided upsets and kept their playoff hopes alive. But in several cases, victory was snatched from the jaws of defeat.

Using Pro Football Reference's Win Probability calculator, one can discover how Miami, Notre Dame, Michigan, and Stanford all defied the odds in season-saving wins. To best enjoy this article, treat the red highlighted links below as footnotes and click away.

Miami's game with Duke may not have had significant implications on the national playoff race, but it featured the largest win probability fluctuations of the night. Everyone knows the game was seemingly over before Miami scored its one in 1000 touchdown. But what fewer people realize is how Duke needed a similarly improbable comeback to go ahead 27-24.

When Duke started on its own 25 down 24-12 with under six minutes left, the Blue Devils had a 0.4% chance to win. Or in other words, a one in 250 chance. The Devils converted on a fourth and 8 during the touchdown drive, but the odds were still in Miami's favor. Even after forcing Miami to punt, Duke only had a 29.7% chance to travel 80 yards in 105 seconds for a touchdown. That number dropped to 27% later in the drive. Yet Miami could not make the defensive stop, and Duke marched in for an apparent game-winning score with six seconds left.

The miracle play helped Miami avoid falling into deeper disarray, spoiled Duke's bid to improve to 7-1, and came with controversy. But years from now, none of that will matter.

Most people first saw the insane Miami lateral play after commercials in the Notre Dame-Temple clash. While that game lacked a truly insane finish, it featured no shortage of late game win probability swings. When Temple began its game-tying drive with 3:04 in the third quarter, the Owls only had an 8.8% chance of winning.

Despite driving the length of the field, the calculator only gave them a 6.3% of victory as they faced fourth and goal from Notre Dame's one with 10:55 left. Jahad Thomas' tying TD run changed things dramatically, improving Temple's win probability by 31.3% on a single play.

Notre Dame's win probability fell below 50% after the Owls edged ahead 20-17 and fell to a game-low 44.2% on a key third down. However, Alize Jones' 45-yard reception with three minutes left put Notre Dame in business, and Temple's win probability never reached 25% after ND wideout Will Fuller hauled in the go-ahead 17-yard score.

As those battles were fought, Michigan and Minnesota simultaneously waged a wild war for the Little Brown Jug. The Gophers dug themselves into a 14-3 second quarter hole, so the probability model gave them just a 3.7% chance to win. Minnesota then rallied with 13 straight points to lead 16-14 at halftime and have a puncher's chance for the upset.

The Gophers played inspired football for recently retired coach Jerry Kill, building a 26-21 lead with nine minutes left. At that point, Minnesota's win probability exceeded 70%. The Wolverines managed to regain possession and drive into Minnesota's red zone, but the model heavily favored Minnesota as the visitors faced a key third and 10. The Wolverines then swung the model completely the opposite direction, scoring the TD and two-point conversion.

Suddenly the Gophers were heavy underdogs again, especially after backing into a seemingly hopeless  third and 17. Minnesota converted with a 17-yard pass, but the Gophers faced another dire situation two minutes later. Somehow, Michigan allowed the drive to continue, and when Minnesota reached the one yard line on another great pass from Mitch Leidner, the model finally favored the Gophers. Yet the Wolverines defied the odds on two straight plays to emerge victorious.

If anyone was in the mood for a tricky finish to a treat-filled evening, the #Pac12AfterDark delivered again with Stanford visiting a surprising Washington State team. For the first 33 minutes, Stanford's offense disappeared against the Cougars' generally mediocre D. Kicker Erik Powell hit five field goals for Washington State, who could have easily routed Stanford if a few scoring drives had ended in touchdowns.

Instead the Cardinals trailed 15-3 and faced slim odds to remain unbeaten in Pac-12 play. The teams traded quick scores, but things still looked dire for Stanford late in the third quarter. Facing a 22-13 deficit, Kevin Hogan boosted Stanford's hopes with a 59-yard TD run. The Cardinal then moved into the model's driver's seat following a Wazzu pick. Stanford took full advantage, going up 27-22 to silence the home crowd.

The Cougars' win probability dropped from 78% to 11.5% in just over three minutes. Many teams would have folded, but Washington State responded with a gutsy go-ahead drive. After forcing a Stanford punt, the Cougs win probability was surprisingly only 39.8%. Clearly the model knew this was Washington State with the ball.

After getting a first down with 3:40 left, the Cougs threw on first down looking for a big play. The pass nearly worked, but Falk's screen pass on second down was intercepted by Stanford. In an instant, the Cardinal went from slim underdogs to heavy favorites once more.

The Cougar defense held Stanford to three points, but the odds were not in Washington State's favor to begin the final drive. The Cardinal allowed no deep balls, leaving Wazzu in this spot with 21 seconds left. Luke Falk delivered a 23-yard strike and clocked the ball with 13 ticks left, but the odds still favored Stanford. After a seven-yard out route, two yard run, and a timeout, kicker Erik Powell had a chance to be the hero from 44 yards. The model only gave Wazzu a 41.6% chance of winning, and Powell's miss prevented the Cougars from clawing back again. Stanford escaped the Palouse by a 30-28 count.

So what does all this analysis tell us? First, each winning team had a win probability below 50% in the final four minutes on action. Three teams won on the final play. And one team won a game that will go down in college football lore.

What makes the weekend particularly incredible is that each of these four games featured a ranked team seeking a 10+ win season. Win-probability analysis is simply a way to contextualize the chaos that will live on long in college football lore.

### Upcoming Games

Sep 26th, 1:00 PM

Indianapolis +3 -115

Tennessee -3 -115

@

Sep 26th, 1:00 PM

LA Chargers +7.5 -107

Kansas City -7.5 -107

@

Sep 26th, 1:00 PM

Washington +7 -102

Buffalo -7 -102

@

Sep 26th, 1:00 PM

Chicago +8.5 -107

Cleveland -8.5 -107

@

Sep 26th, 1:00 PM

Atlanta +2.5 -107

NY Giants -2.5 -107

@

Sep 26th, 1:00 PM

Cincinnati +6.5 -107

Pittsburgh -6.5 -107

@

Sep 26th, 1:00 PM

Baltimore -7.5 -110

Detroit +7.5 -110

@

Sep 26th, 1:00 PM

New Orleans +3 -110

New England -3 -110

@

Sep 26th, 1:00 PM

Arizona -6 -110

Jacksonville +6 -110

@

Sep 26th, 4:05 PM

Miami +1 -107

Las Vegas -1 -107

@

Sep 26th, 4:05 PM

NY Jets +7.5 -107

Denver -7.5 -107

@

Sep 26th, 4:25 PM

Tampa Bay +2.5 -110

LA Rams -2.5 -110

@

Sep 26th, 4:25 PM

Seattle -2.5 -107

Minnesota +2.5 -107

@

Sep 26th, 8:20 PM

Green Bay +3 -107

San Francisco -3 -107

@

Sep 27th, 8:15 PM

Dallas -3.5 -110

@

Sep 30th, 8:20 PM

Jacksonville +0.5 -116

Cincinnati -0.5 -116

@

Oct 3rd, 1:00 PM

Detroit +6 -110

Chicago -6 -110

@

Oct 3rd, 1:00 PM

Carolina +5.5 -110

Dallas -5.5 -110

@

Oct 3rd, 1:00 PM

NY Giants +6.5 -110

New Orleans -6.5 -110

@

Oct 3rd, 1:00 PM

Cleveland -1 -110

Minnesota +1 -110

@

Oct 3rd, 1:00 PM

Houston +16.5 -110

Buffalo -16.5 -110

@

Oct 3rd, 1:00 PM

Tennessee -6.5 -110

NY Jets +6.5 -110

@

Oct 3rd, 1:00 PM

Kansas City -6.5 -110

@

Oct 3rd, 1:00 PM

Indianapolis +2.5 -110

Miami -2.5 -110

@

Oct 3rd, 1:00 PM

Washington -2.5 -110

Atlanta +2.5 -110

@

Oct 3rd, 4:05 PM

Seattle +3.5 -110

San Francisco -3.5 -110

@

Oct 3rd, 4:05 PM

Arizona +6 -110

LA Rams -6 -110

@

Oct 3rd, 4:25 PM

Baltimore -1.5 -110

Denver +1.5 -110

@

Oct 3rd, 4:25 PM

Pittsburgh +6.5 -110

Green Bay -6.5 -110

@

Oct 3rd, 8:20 PM

Tampa Bay -3.5 -123

New England +3.5 -123

@