The College Football Playoff kicks off on New Year’s Eve with Washington vs. Alabama and Ohio State vs. Clemson. Alabama currently is a 14.5-point favorite, while the Buckeyes are favored by 3.
After a 13-game season (12 for Ohio State), there is plenty of data to compare these four teams. The question is whether those statistics agree with Vegas and conventional wisdom.
College Football Playoff: Alabama (-14.5) vs. Washington
Conventional wisdom for this game is that the Tide will roll. ESPNs Mark Schlabach wrote the following:
“We can try to pretend the Peach Bowl will be close because we want to watch a competitive game. It won’t be. Tide roll. Alabama 27, Washington 10.”
But after looking at the statistics, I’m not so sure. There are three statistics that are accepted as predictive when it comes to understanding how a team has performed (these ideas been have been emphasized repeatedly by ESPNs Bill Barnwell and applied to his excellent NFL analysis).
- Record in close games (decided by 7 points or less)
- Fumble luck
- Predicted record by Pythagorean expectation
After an analysis of the previous 10 years of college football, I also have found that there is a correlation between winning percentage and quarterback rating (R2 = 0.60).
The rationale of looking at these statistics is that there is enough luck in a 13-game season that there is an extremely small difference between going 13-0 and 12-1. Thus, just looking at a team’s record does not give a complete picture about how good it has or hasn’t been.
There is no real difference between Alabama and Washington when it comes to records in close games (1-0 for Alabama, 2-0 for Washington) or fumble luck (both recovered 52 percent of fumble opportunities). The difference between the two teams shows up in their point differentials.
The Pythagorean expectation — developed by famed Sabermatrician Bill James — uses point differential to predict how many games a team “should” have won. This is a quantification of a fairly intuitive idea: A team that wins via blowout is better than a team that wins lots of close games. And a team that loses a lot of close games probably is better than its record indicates.
I took that one step further in this particular analysis and further divided the data into home vs. road/neutral against Top 25 opponents (as ranked by the ESPN Football Power Index at the end of the regular season).
The rationale is twofold. First, teams don’t play cupcakes on the road and the playoff games will be played at a neutral site, so it makes sense to see if a team plays better or worse on the road. Second, how a team performs against top competition should be predictive of whether it is elite or merely feasting on a relatively easy schedule.
Overall, Alabama outscored its opponents by almost 29 points per game while Washington “only” outscored its opponents by more than 27. This small difference in scoring margin makes a significant difference in the predicted win total.
But a closer inspection of the data shows that Alabama was historically dominant at home, posting a scoring margin of 32.9. That margin shrinks considerably on the road (22.2). Most teams play worse on the road, but Washington was not one of those teams. The Huskies posted similar overall (27.2), home (27.6) and road/neutral (26.8) point differentials.
What this means is that at a neutral-site game (such as the CFP), the predicted win totals based on point differential favor Washington (10.9), not the Crimson Tide (10.6).
Alabama has been more dominant against Top 25 teams than Washington. Indeed, the Tide dominated a USC team that beat the Huskies. But many of those top teams that Alabama played were SEC teams that are ranked higher by FPI than their records would indicate, like LSU (ranked No. 7). Also, USC changed QBs between playing Alabama and Washington.
One reason that teams excel on the road is quarterback play. And this is where Washington again has a distinct advantage.
Alabama true freshman QB Jalen Hurts has been fantastic this season, but like most true freshmen (and most QBs in general), he has played much better at home than on the road. Conversely, Washington QB Jake Browning has played better than Hurts overall and way better on the road.
In 2015, Browning compiled a home QB rating of 154.5, fairly close to this year’s rating. However, his road-neutral rating was more than 30 points lower (122.8). The explanation for Washington’s 12-1 record this season and 7-5 last season is tied almost entirely to improved quarterback performance on the road.
The SEC Championship Game was a prime example of how Washington might exploit this. Hurts looked lost in the first quarter, as Florida held the Alabama offense to minus 7 yards on 7 plays. Unfortunately for the Gators, the Tide also scored 16 points on defense and special teams.
But the blueprint for Washington is the same. Florida forced Hurts to make plays early and he was unable to do so. Because the talent differential between those two teams — particularly Florida’s offense against Alabama’s defense — was so wide, he never had to.
If Washington can jump ahead early on the back of fantastic play from Browning and force Alabama to play from behind, I’m not sure it is capable of doing it. Regardless, the statistics say that 14 points is way more credit than the Crimson Tide should be getting.
I think Washington pulls the upset, 31-27.
College Football Playoff: Clemson vs. Ohio State (-3)
This game seems like it will be the better of the two playoff games. Going to Schlabach’s preview again, he says, “It’s going to be a classic. Clemson 42, Ohio State 40.”
Again, based on the statistics, I don’t think that’s the case.
First, Clemson has been incredibly fortunate. The Tigers are 6-1 in close games this season. This includes a home loss to Pitt and a home victory against North Carolina State in overtime. Typically, teams win close games about half the time, so the Tigers are not as good as their 12-1 record indicates.
Ohio State isn’t much better in this department. The Buckeyes are 4-1 in close games. Ohio State has been fortunate, particularly in a 1-point escape against Michigan State in East Lansing.
Fumble recovery percentage is random. Some of the reason for the number of close games for Clemson could be attributed to fumble luck. The Tigers have recovered only 35 percent of fumbles this season, compared to 55 percent for Ohio State.
A good example is the aforementioned North Carolina State game. Three Clemson drives ended in Wolfpack territory after fumbles. With a little bit more luck, the outcome would have been lopsided.
Looking at point differential shows the real difference between these two teams.
Ohio State has a significantly better overall and home point differential and predicted record. The road-neutral scoring margin shows teams with similar profiles.
But Clemson has struggled against better teams. All four of its games against Top 25 opponents ended with one-possession margins. Compared to Alabama and Washington, Ohio State has struggled as well, but not as much as Clemson. (See: Ohio State’s 45-24 win vs. Oklahoma in Norman.)
The QB comparison also shows some differences, but nothing game-changing.
Clemson QB Deshaun Watson has performed better overall, but Ohio State’s J.T. Barrett has been more consistent. What this means is that if you think that the road-neutral ranking up above is predictive of how the QBs will play in the playoff game (I do), their throwing skills are a wash.
And while Watson is known for running the ball, he only averaged 40.7 rushing yards per game this season. Conversely, Barrett averaged 70.6 rushing yards per game. This was an enormous weapon in close games against Michigan (125 yards) and Wisconsin (92 yards). Indeed, during Ohio State’s lone loss against Penn State, the Nittany Lions shut down Barrett to his fewest rushing yards of the season (26).
To be fair, Clemson didn’t ask Watson to run as often. And both QBs averaged more than 4 yards per carry in 2016.
These two teams are closely matched. I wouldn’t be surprised to see a close game. But Clemson has been an incomplete team this year that has relied on Watson to keep the Tigers in the game. Conversely, Ohio State has received consistent QB play but nothing spectacular, which I think indicates that it is a more balanced team.
I can’t ignore the huge difference in overall point differential between the two teams. Ohio State’s differential is that of a team a full win better than Clemson. Clemson will have to play a perfect game to keep this one close. And I don’t see it.
Ohio State wins, 31-17.