FanPost

ESPN's FPI Projections for the Pac12 Teams

I think that many Husky fans have seen ESPN's FPI projections for 2021. I know that a lot of people dismiss any projections by ESPN, but their FPI is less opinion and more data-driven. In the past, it has been fairly accurate. If they say that one team has a 70% chance of winning, it generally means that teams with that win percentage will win about 7 of the 10 games. That gives them some ‘wiggle room'; they aren't predicting who will win a particular game, just which team looks to have the better chance. If they are wrong, then it was just one of the less-likely outcomes.

There are a couple of ways that they have of looking at the wins for the season by a particular team. One way that usually jumps out to people is to look at which games the team is favored in (e.g. a greater than 50% chance of winning). The other is that they project an overall W-L record by combining the win percentages of all of a team's games. For example, if a team is favored in all of their games, but with a win percentage of 51% in each game, chances are good that they are more likely to win 6 or 7 games than all 12 games.

Here is a summary of the Pac12 team's projections for the year. I've included both the number of games each team is favored in as well as ESPN's FPI's projections for number of wins.

Team

Favored

Projected Wins

Arizona

1

3.3

Arizona State

9

8.3

California

6

5.1

Colorado

4

5.3

Oregon

11

9.5

Oregon State

2

4.4

Stanford

6

6

UCLA

5

6.1

USC

10

8.6

Utah

9

7.9

Washington

10

8.5

Washington State

4

5.3

As you can see, Arizona is projected to win some games that it is currently not favored to win while teams like Oregon (and UW) are expected to lose games that they are favored in. Oregon State has the biggest difference; they are expected to win 2.4 more games than they are favored in. (One of the reasons that I think some people don't like the ‘Projected Wins' is the use of fractional values for wins. It just looks odd to talk about "0.4" wins. I like to think of it this way: they are more likely to win 4 games than 5. Similarly, Utah is more likely to win 8 games than 7 with that projected win total of 7.9.

In most cases you can determine the projected wins by adding up the win percentage of all of the games-or at least get very close. (Not sure why it works for some, but not others.) For example, adding them up for UW gets 8.4 wins while ESPN's FPI has 8.5.

Some Observations from the Games

Oregon has a 99.2% chance to beat Stony Brook; that's the highest win percentage of any Pac12 team in any game. (I can't argue too much about that.) Arizona State's 99% chance of beating Northern Arizona is a close second.

Arizona has only a 5% chance of beating Oregon; that's the lowest win percentage of any Pac12 team in any game. Arizona also has only an 8.3% chance of beating USC; that's the second-lowest win percentage. The only other teams with a less than 10% chance of winning are Cal (9.8% vs Oregon) and Oregon State (8.7 % vs Oregon).

Here's another way to look at the data. I've bracketed the win percentages in intervals of 10% (0-10%, 10-20%, etc.) to see the distribution for the teams.

<10

10-20

20-30

30-40

40-50

50-60

60-70

70-80

80-90

>90

Arizona

2

5

1

1

2

0

0

0

0

1

Arizona State

0

1

0

1

1

2

0

3

2

2

California

1

2

3

0

0

3

2

0

0

1

Colorado

0

2

2

3

1

1

0

1

1

1

Oregon

0

1

0

0

0

2

1

1

2

5

Oregon State

1

2

3

4

0

0

0

1

0

1

Stanford

0

0

3

1

2

3

1

2

0

0

UCLA

0

0

4

1

2

0

1

3

1

0

USC

0

0

1

0

1

0

3

4

1

2

Utah

0

0

0

1

2

2

1

3

2

1

Washington

0

0

0

1

1

1

2

2

4

1

Washington State

0

4

1

1

2

0

1

1

1

1

Oregon is a heavy favorite (80% or more) in 7 of their games. Stanford isn't favored in any of their games by more than 80%. UW and Utah are two teams that are favored by at least a 30% or more in all of their games. (UW's lowest is vs Michigan at 38.6% while Utah's lowest is 38.2% vs USC.) UCLA has at least a 20% chance of winning in each of their games, but between 20% and 30% in 4 of their games (UW, Oregon, Utah, and USC). Stanford is an example of being in the middle; they don't have a win percentage above 80% in any game, but they don't have a win percentage below 20% in any game.

ESPN adjusts the win percentages each week based on the data from the games. (The preseason data is based on last year, returning production, and recruiting projections.) Most of the time the changes are small week-to-week, but occasionally there will be some big changes if there are, for example, upsets early in the season. A win by Colorado in week 2 vs Texas A&M, for example, would mean a big jump in Colorado's win percentages for many of their games. I'm hoping to keep track of the win percentages each week; I'll be watching for big changes in win percentage as well as total wins. If you want to check out UW's win percentages each week, here is the link to the site: http://www.espn.com/college-football/team/fpi/_/id/264