Pac-12 Rankings and Projections Post Week 4

This week I’m taking a little different look at the rankings of the Pac-12 teams. Now that we have a few games in, I’m going to look more at the analytics than the human polls-at least partly because most Husky fans are already aware of where UW and the other Pac-12 teams are in the polls, but also because those analytics are starting to have enough data from this year that we can start to get a clearer picture.

Rankings Composite

A summary of many different rankings for all FBS teams can be found here: https://masseyratings.com/cf/compare.htm

These rankings include some human polls (like the AP and Coaches polls), but mostly are different ‘computer’ rankings-in other words, rankings based on analyzing the performance of the teams and not as much on human perception. Different people have come up with different formulas for rankings, which is why there are so many and why the rankings can differ so greatly for the same team.

Among those rankings, UW is as high as #5 in the RAMS (Record, Average, Margin, and Schedule) Rating System, and as low as #65 in the Logan’s Sports Rating (where they are still behind Michigan State). By averaging all of the (currently 62) different rankings, UW is #21. Here is the data for all of the Pac-12 teams.

The way this works is that each team’s average ranking (‘Mean’) is calculated from all of the different rankings. The team with the lowest Mean is ranked first, second-lowest is second, and so on. For those that don’t remember statistics, the ‘Median’ is the middle value; half of the values are above and half are below. In general, large differences between the ‘Mean’ and the ‘Median’ indicate that there are some outliers in one direction. Utah is a good example of this. Their ‘Mean’ is lowered because there are a few rankings that have them below 50 (like the one at 79) while the majority are above 20. The Standard Deviation (‘Std. Dev.’) is another way of looking at the variability for a team; the higher the value, the more variability there is between all of the various rankings.

There are honestly some rankings that you have to scratch your head (to put it politely) and wonder if they are taking all of the games into account. For example, the LAZ Index has Oklahoma at #4, Arizona State at #31, and 0-4 Colorado above 4-0 Kansas. But with 59 different rankings being used, ones like that which are clearly outside the norm don’t significantly impact the overall composite. In fact, if a human poll came out with a Pac-12 Power Ranking with teams in the order in that table, I don’t think that anyone would have a lot of problems with it. (I’m sure that there are many Husky fans that believe that UW should be above Oregon, but that’s a separate discussion.) So, while any individual ranking may not be good, the composite of all of them seems to produce a decent result.

If you compare this ranking with the AP and Coaches polls, there are a few teams that are above UW in this ranking but not in either poll. Those teams are Minnesota (which just moved into the AP top-25), Oklahoma (which just dropped below UW in the top-25), Cincinnati (getting votes), Florida State (which just moved into the AP top-25), and LSU. I believe that some of the rankings are using results from last year because there are still so few games this year, so I expect that there will be fewer cases like this as more games get played (they get more data).

Along that line, it is helpful to point out that UW started out at #67 in the preseason composite ranking. Moving from #67 to #21 in just a few weeks is a great accomplishment. At a minimum, I would expect the variability for UW to be reduced in the coming weeks as long as they keep winning, even if their ranking does not continue to rise as much as it has.

The following table shows how the composite rankings for the Pac-12 teams has changed so far this year.

As you can see, UW has made the biggest jump of all Pac-12 teams, although USC is close behind. Remarkably, Stanford has not changed much at all since the preseason; even week-to-week there has not been much change. Not surprisingly, Colorado and Arizona State have had the biggest falls. Utah is the only other Pac-12 team to fall, although not by as much. The rest of the teams have all improved-and some, like UW and Cal, have improved every week.

Advanced Stats

Included in those rankings are some of the more popular advanced-stats-based rankings. These include the S&P+, FEI, and ESPN’s FPI. Another one that I look at isn’t quite as popular: F+. The following table shows where each Pac-12 is ranked in each of them.

For most of the teams there is a lot of consistency between those rankings and between these rankings and the composite rankings above. One exception is Arizona State whose composite ranking (94) is well below those advanced stats rankings. That may well indicate the difference between how well they should be playing and how well they are playing.

ESPN FPI

Here is the updated information for the Pac-12 teams including a comparison to their preseason ranking and win projections.

It appears that the Pac-12 should have at least 6 teams in bowl games, and could have as many as 8. But, there are a lot of games yet to be played. California, for example, still has to play Washington, Oregon, USC, Oregon State, and UCLA. WSU still has to play USC, Oregon State, Utah, and Washington. That makes this weekend’s game between those two teams critical because the loser could have a much harder time getting to 6 wins. Each is currently favored in only two more games this year after this weekend. (WSU is a slight favorite this weekend.)

ESPN Game Projections

In addition to the updated rankings and Win-Loss projections, ESPN continues to update the win percentages for the remaining games. Here are the number of conference wins expected for each Pac-12 team (current conference wins plus number of games favored).

• Arizona (0+2) 2
• Arizona State (0+1) 1
• California (1+2) 3
• Colorado (0+0) 0
• Oregon (1+7) 8
• Oregon State (0+5) 5
• Stanford (0+2) 2
• UCLA (1+4) 5
• USC (2+6) 8
• Utah (1+8) 9
• Washington (1+7) 8
• WSU (0+3) 3

In the past there have been plenty of upsets in the Pac-12, so I don’t expect that those will be the final results. But it is helpful to get a picture of what is expected to happen-especially if you want to compare it to preseason expectations.

Here are their projections for all of UW’s remaining games:

UW is currently favored in every game except the game at Oregon-although the win percentage for UW continues to improve. The week-to-week changes are much smaller now than they were earlier in the season. We still may see some big changes later on, but I expect most will be in this range week-to-week.

Here are all of the winning percentages for this week’s Pac-12 games.

As you can see, the home team is favored in every game except the UW-UCLA game. Two games appear to be close while the others appear that they could be one-sided.

Should we believe ESPN’s FPI? If you had used ESPN’s FPI each week to pick the winners of all of the Pac-12 games so far this year, you would have been correct for 88% of them. The games which they missed were: Arizona over San Diego State, Oregon State over Boise State, WSU over Wisconsin, Eastern Michigan over Arizona State, and UW over Michigan State. The WSU and Eastern Michigan wins were clear upsets that few saw coming. The other games were either an over-rating of a team based on last year, an under-rating of a team from last year, or both. Still, that isn’t a terrible result.

Talent Comparison

I’ve shown before that the team with more talent (based on the 247Sports ‘College Team Talent’) has a good chance of winning-anywhere from 51% to 94% depending on the difference in talent and which team is the home team. If you had used that to pick the winners of the Pac-12 games, you would have been correct for 93% of the games so far this year. The only ‘upsets’ this season have been Air Force beating Colorado, WSU beating Wisconsin, and Eastern Michigan beating Arizona State. If you use that for this week’s games, it would be:

• California (653.31) over WSU (571.02)
• Utah (705.81) over Oregon State (637.52)
• Arizona (652.06) over Colorado (616.51)
• USC (860.12) over Arizona State (666.09)
• Oregon (877.87) over Stanford (757.81)
• UW (761.72) over UCLA (732.85)

This week those projections match up with the ESPN FPI projections. But if you continue to use that ‘College Team Talent’ ranking for the remaining games, you will likely see some more upsets-especially in weeks 11 and 12. (I’ll leave it to you to figure out which games could be upsets based on this.)