Every college football season produces a heaping portion of what-if. That’s the nature of any sport, especially one that relies on 12-15 games to make all of its decisions, from title rings to firings and everything in between.
After simulating 2017’s college football season in retrospect, it’s clear Washington missed a huge opportunity
Based on how teams played in each game, let’s try and figure out how the season was likeliest to go.
Photo by Otto Greule Jr/Getty Images
2017’s College Football Playoff race was determined by an illegal shift, a one-handed catch, a comeback fallen just short, and a blowout in Iowa City. We spend an interminable offseason talking ourselves through every minuscule in-and-out, and a couple of plays end up making a huge difference.
While we’re talking about randomness, however, let’s see what the numbers have to say about what could have, or should have, happened this fall.
To do that, let’s play with postgame win expectancy.
In my Football Study Hall stat profiles, you’ll find that number for every game that’s been played. Postgame win expectancy is intended to say, “Given your success rates, big plays, field position, turnovers, etc., you could have expected to win this game X percent of the time.”
It treats certain other factors as outside a team’s control. For example, a team’s fumble recovery rate will usually end up around 50 percent. If you recovered all five of a game’s fumbles, that’s hard to replicate, so the simulation will try to see what happened had the ball bounced more normally.
Postgame win expectancy has nothing to do with pre-game projections or opponent adjustments. It simply looks at how you played and makes a retroactive forecast, so to speak.
I used these postgame figures to simulate the 2017 season a thousand times. I wanted to see what was most likely to have occurred this fall, based on how the games were played and how they could have ended.
In 49 percent of the simulations, Washington finished 11-1 or better.
Chris Petersen’s Huskies went 10-2, with 10 mostly certain wins and two tossup losses. In only one of their 10 wins was their postgame win expectancy below 92 percent. Strangely enough, that one game was against Rutgers.
Based on in-game stats, they’d have beaten Arizona State (to whom they fell, 13-7, after two short missed field goals) 48 percent of the time, and they’d have beaten Stanford 40 percent of the time. The odds of them losing both of those games based on the numbers those games produced: only about 31 percent. The odds of them winning both were 19 percent.
Granted, even if they’d have won, they’d have still had to get past USC in the Pac-12 title game to reach the Playoff. S&P+, not a big fan of the Trojans, would have given UW about a 70 percent chance in that game.
Out of hundreds of 2017 college football seasons, you could say Washington makes the Playoff at least one-third of the time. Reality didn’t produce one of those times.
Next on the Unfortunate List: TCU (31 percent of going 11-1 or better), Penn State (25 percent), USC (19 percent), and Auburn (18 percent).
In 64 percent of re-simulations, at least one of these four teams finished 11-1 or 12-0 before conference title games. In reality’s simulation, none did. USC lost to Washington State despite 76 percent win expectancy, TCU lost to Iowa State at 64 percent, Penn State lost to Michigan State at 40 percent, and Auburn lost to LSU at 25 percent.
(Ohio State’s win expectancy was nearly 100 percent in almost every game, but two games at 4 percent — Oklahoma and Iowa — kept it out of the Playoff.)
Who was the luckiest from a stat perspective? Bama.
According to postgame win expectancy, there was a 40 percent chance of Alabama finishing with at least two losses. Clemson had a 39 percent chance, while Oklahoma (21 percent) and Georgia (19 percent) were a little more secure.
The Tide’s definitive loss to Auburn (with how the game’s basics played out on the field, the Tide would’ve only won 5 percent of the time, basically needing every bounce of the ball to go their way), combined with two not-completely-certain wins (83 percent against Texas A&M, 81 percent against Mississippi State), meant they were pretty close to the danger zone. Clemson played two straight tossup games: a narrow loss to Syracuse (51 percent) and narrow win over NC State (47 percent).
Oklahoma finished with two less-than-certain outcomes, losing to Iowa State (64 percent postgame win expectancy) and beating Kansas State (62 percent), but there was a much greater chance of the Sooners winning both (40 percent) than losing both (14 percent).
Georgia had one absolute loss (1 percent in the first game against Auburn), but none of the Dawgs’ wins, even the one-pointer over Notre Dame, had an expectancy below 88 percent.
College football is never going to be rational, but in the least irrational version of 2017, we probably end up with Washington instead of Alabama.
Finally, the plucky Crimson Tide caught a break.
See More:
