With Questions Looming
Photo Credit: Eric Francis

Hot Reads: About Those 5.5 Wins

April 06, 2017

ESPN’s Football Power Index isn’t very high on the Huskers in 2017. We knew that last week thanks to some handy analysis from MGoBlog,  but it seemed to gain traction in Nebraska yesterday when ESPN released its own projected wins and losses.

That showed a projected record for Nebraska of 5.5-6.5 in 2017. (MGoBlog’s estimated FPI record had Nebraska at 5.57 wins. Not bad.) That got me to wondering, in a very general sense, how accurate ESPN’s FPI rankings have been over the years.

Here’s Nebraska projected record by FPI for the previous three seasons (Year: Preseason FPI Rk | Projected Record | Actual Record | Final FPI Rk):

2016: 25 | 8.9-3.5 | 9-3 | 48
2015: 43 | 7.7-4.5 | 5-7 | 39
2014: 41 | 7.7-4.4 | 9-3 | 28

Super small sample size there, but the absolute mean error for wins is 1.3 wins. With more time I’ll go back and do this for all 128 teams over three years and see what that difference is over that span just because I now want to know what it is. It’s worth noting, however, that FPI’s game predictions have ranked in the top 10 here the last two years (the only two years it was included).

I did do that three-year, all-teams check for the FPI preseason to final rankings. How close to ESPN’s preseason rank were teams finishing in the same ranking system?

You can say this for FPI: It remains pretty consistent year to year. The absolute mean difference in 2014, 2015 and 2016 fell between 13.15 and 13.92 each year. The standard deviation ranged from 17.05 to 17.92. Take all three years as one set of projections and you have an absolute mean error of 13.55 and a standard deviation of 17.37.

Yeah, I know, boring right. What that means is that over the last three seasons, on average a team’s preseason ranking has been off by 13.55 spots. Assuming those numbers remain consistent, the most likely final ranking for Nebraska in FPI falls somewhere between about 40th (Temple, BYU and Memphis ranked 40th the last three years) and 75th (Colorado State, Vanderbilt, Colorado).

Of course that’s not really the company anyone affiliated with Nebraska football — fans, players, coaches, administration — wants to keep. But it’s projections like this that continue to make me think the key question for the Huskers in 2017 becomes how they outperform this preseason expectation (and not “this projection is clearly wrong.”)

That’s what it’s going to take because I think you can put some stock in this stuff. And that sort of thing happens. In 2016 Colorado was 59th in preseason FPI and finished 40 spots higher. Kansas State went from 57th to 30th last year. Of course there are similar differences on the other end.

But those are the outliers and that’s probably the main takeaway here. If you put any stock in things like FPI, it reframes what, say, a 9-win season for Nebraska in 2017 would be — unlikely, but of course not impossible.

And that’s part of cognitive dissonance here and with any other set of rankings. A fan base pays attention to one projection. It’s happening in Nebraska right now and every other place that bothers to be interested in how its team is being perceived. But we have to be careful not to judge the validity of rankings based on how close the one projection we pay the most attention to is. Say Nebraska wins nine games this season. It will have outperformed its FPI projection, and some people will always use that as a reason to dismiss these things out of hand. But what if, say, 70 percent of the remaining 127 projections were mostly right? Does that have value? I think it does.

That said, I still feel like 5.5 wins is a little low for Nebraska, but I can see why it’s in that range based on how the Huskers look from a numbers perspective. Guess we’ll find out if it is low in about eight months.

The Grab Bag

Today’s Song of Today

https://youtu.be/FXl6D4oYIeg

  • Never miss the latest news from Hail Varsity!

    Join our free email list by signing up below.

Share via
Copy link
Powered by Social Snap