Hot Reads: So You're Saying There's a 6% Chance
Photo Credit: Eric Francis

Hot Reads: So You’re Saying There’s a 6% Chance

July 09, 2019

I can stop estimating now.

ESPN released, with little fanfare, its updated FPI ratings for the the 2019 college football season. This round included game-by-game projections for every team, with win probabilities (which produces a win total), and conference-title odds. That's a good amount of new information.

Back in May, I tried to project what that win total, and those win probabilities, would be based on the FPI 1.0 ratings:

In case you weren't adding those percentages up as you went along, they total . . . extended drumroll . . . 7.81 wins. Barring any big changes from the FPI 1.0 ratings, I'm guessing the projected record there will be pretty close to 8-4 when ESPN gets around to releasing it.

There were some slight changes to the ratings. Nebraska, for example, was viewed as 8.0 points better than the average team in version 1.0, but is at 8.2 now. Factor in similar increases and decreases for other teams and the new FPI has the Huskers' projected record at 8.1-4.2. Despite using a generic 2.5-point adjustment for home field––I'm assuming FPI has something more specific––the projected win probabilities by game didn't end up being too far off in most cases. (Holy Wisconsin, though.)

GAME PROJ. WIN PROB. ACT. WIN PROB.
S. Alabama 99% 98%
at Colorado 65% 64%
N. Illinois 96% 91%
at Illinois 75% 75%
Ohio State 39% 42%
Northwestern 73% 73%
at Minnesota 38% 40%
Indiana 66% 72%
at Purdue 65% 64%
Wisconsin 50% 66%
at Maryland 65% 64%
Iowa 50% 51%

FPI gives Nebraska a 6% chance to win the Big Ten. That ranks seventh in the conference and is sort of a cutoff point based on the percentages. There's a steep drop to the eighth-ranked team, Wisconsin.

TEAM CHANCE OF WINNING CONF.
Michigan 48.2%
Penn State 10.4%
Ohio State 8.7%
Michigan State 8.3%
Iowa 8.0%
Minnesota 7.9%
Nebraska 6.0%
Wisconsin 1.4%
Northwestern 0.6%
Indiana 0.3%
Purdue 0.2%
Maryland 0.0%
Illinois 0.0%
Rutgers 0.0%

If you're interested in what goes into FPI, there's a good explainer here. That offers some insight into some of the questions you might have based on these ratings. Or at least some of the questions I have, like . . . 

Why is FPI down on Wisconsin? The Badgers have an average number of returning starters coming off a below-standard season and they don't get full credit for a returning starter at quarterback. It also has a somewhat difficult schedule (28th nationally). That’s my best guess anyway.

Why is FPI down on Ohio State? The Buckeyes get some transfer-quarterback points, but not returning-starter points from Justin Fields. Also, coaching changes are treated as regression to the mean for FPI's purposes. If that seems like a basic way to handle that, I would argue that it's at least elegantly basic. If a team fires its head coach for performance, it is hoping to upgrade and if it succeeds in upgrading that team should improve, getting closer to whatever the mean is for that program. If a team loses a coach it didn't want to leave––I don't know, let's just use Urban Meyer as an example––the next guy is likely to have a hard time replicating such lofty results. It's nothing against the new guy, just a matter of probability. I kind of love this simple solution to an interesting problem: How do you account for coaching changes in a preseason rating? And I'm trusting that ESPN's crew did the modeling to make sure it makes sense. Point is, Ryan Day could be very good and still not match Meyer's excellent results and FPI has a way to account for that.

Why is FPI high on Penn State? I'm not sure. I fear I have overvalued the Nittany Lions a bit myself this offseason. I had Penn State fourth in my Big Ten power rankings for the 2019 Yearbook and I already feel like that was too high. There's just more uncertainty with Penn State than there is Michigan State, though the Lions' have the potential to produce a pretty nasty defense.

If we played this 2019 season 100 times, is Michigan really winning the Big Ten 48 times? That's what FPI is saying, and I do think the Wolverines are a deserving favorite. While the gap between Michigan and the next-best team looks huge in terms of conference-title odds, it's actually pretty normal for a Power 5 favorite over the last six seasons. Since 2014, 19 favorites in a P5 conference have had between a 25 and 49% chance to win their conference per FPI. The real outliers are the super favorites. Ohio State was given a better than 68% chance to win the Big Ten in 2015 and 2017. Clemson is given an 88.2% chance to win the ACC this year. Oklahoma has been between 54 and 71% the last four years in the Big 12. Comparatively speaking, the Big Ten is pretty open in 2019 even with Michigan at 48.2%.

The Grab Bag

Today’s Song of Today

  • Never miss the latest news from Hail Varsity!

    Join our free email list by signing up below.

Hurrdat Media Question Marks podcast ad 300 x 600

Hail Varsity March 2023 Cover

Never Miss Another Issue