I love me my stats, and advanced metrics in football have been developing at an incredible rate over the last few years. Last week I posted a little something about DPE, and how it showed that we were actually capable of standing toe-to-toe with Stanford. That worked out pretty well for us, huh?
In light of how illuminating that analysis proved to be, I'm going to take a look at several advanced metrics each week (click on the clicky bit to get taken to the stat source).
1. FEI- The "Fremeau Efficiency Index" attempts to perform a results-independent analysis of a team's FBS games. It doesn't count FCS opponents. It adjusts for the strength of each opponent, and the rankings heavily favor teams that have played tough games. It doesn't care who won or lost the game, although the efficiency with which points were scored are a major factor, and the team who wins usually performs better under those metrics.
2. S&P+- This statistic simply takes another approach, but tries to do essentially the same thing as FEI. It seems to weight play success (typically measured by a % of remaining yardage to a first down) a little more. Also, it purports to take strength of schedule into account, but the top rankings don't reflect that (1. FSU 2. L'ville 3. Baylor).
3. DPE- Drive point efficiency, last week's star statistic, is an opponent-independent metric that capture how much better or worse than average a team is at scoring points and preventing the other team from scoring points, given their field position.
4. CFRC- The College Football Ranking Composite has basically all of the computerized rankings, and aggregates them to tell you who the nation's computers think is the best team. Given the closed room that is the rankings formulas, it seems best to aggregate as many computers as possible to balance out their sometimes significant inadequacies.
Now, the stats-
FEI- Utah is an absolute FEI star. We exploded up the FEI ladder after dropping Stanford, moving from 16th in the nation to 4th in the nation. Missouri, Alabama, and Oregon are the only teams ahead of the Utes. The Utes have the 4th best offensive FEI, the 47th best defense, and the 45th best special teams.
Arizona has plummeted, from a rosy, competitive 18th to 38th. FEI ranks them as having the 22nd best offense, the 31st best defense, and the 81st best special teams. It's that special teams number that pulls them down.
Utah has a major edge according to this metric, and those predicting 12-21 point victories can hang their hat on FEI.
S&P+- Utah is respected by this metric, but we didn't see the same jump as FEI gave us. The Utes went from 25th to 21st last week. We are 14th on offense and 36th on defense.
Arizona was ahead of us here, too but has dropped behind us after a tough loss to USC. They dropped from 12th to 25th in the rankings and are 32nd on offense and 21st on defense.
These metrics make it look like a slight Utah lean, and the home field advantage might be enough to cancel it out. Those of us expecting a 1 score final can point to S&P+ as evidence.
DPE- The Wildcats have a total DPE of .884 (o .470 d -.414). The Utes total DPE is .760 (o .624 d -.136). This stat suggest that the Wildcats are considerably better on defense than the Utes, and would overall be slight favorites to win on a neutral field. This stat justifies the Las Vegas line.
CFRC- the computer aggregate likes the Utes as the 21st team overall. Not counting FEI or S&P+, the Utes' best ranking is 10 which appears in two spots, and worst ranking is 51, which appears once. Last week, the Utes were ranked 37th overall.
Arizona gets its hardest hit of the week from the computer rankings. The CFRC puts them as the 41st team in football, down from 26th last week. The nicest computer ranks them 9th, while the worst puts them at 68th.
On average, the computers think Utah is significantly better than Arizona, and thinks that their resume is superior. These ratings are more about resume than predictions, but would also indicate a likely Utah win.
Overall, the stats seem to trend Utah's way, with DPE as kind of an outlier. One thing DPE doesn't take into account that the other stats do is strength of schedule. Normally, I only think strength of schedule is relevant for resume building, but in this case there is a huge disparity. Sagarin has the Utes' SOS 9 and the 'cats 71. FEI also has a strength of schedule component, and has the Utes' schedule to date as the 5th most difficult, while the Wildcats have the 55th.
No matter how you slice it, every metric agrees that the Utes can score a ton of points, and is the best offense the Wildcats have faced. Those of us thinking the Utes might score 40+ may just be on to something. This principle is apparent by eyeballing the roster and scheme, too: it's hard to imagine the Wildcats' 3 man front being able to stop two guards and a fullback from making a hole for Bubba Poole. The safeties will have to come down, and Dres will beat a corner one on one over the top. 3 stack defenses are designed to conceal weaknesses and stymie spread offenses; a power run game tends to split them right down the middle.
Taking those factors into consideration, the Utes look like the clear favorite according to the advanced numbers.
BYU is 13th in FEI and 19th in S&P+, with a tDPE of 0.522 and a CFRC ranking of 26. Good team.
FEI shows the Pac 12 a ton of love. We have Oregon (1), Utah (4), Stanford (5), UCLA (6), Washington (14), and Oregon State (15). The SEC has 7 teams, but doesn't crowd the top 15 like that.
FEI keeps track of expected points lost through turnovers. In both of the Utes' losses, those expected points are larger than the eventual margin of victory. We already knew this, but those picks cost us the games.
The entire P-12 tended to get closer to the mean DPE. This is expected, and demonstrates the statistical principal of regression. It's more likely that your best performance is an outlier than a representation of what you are typically capable of. This is particularly true in early season CFB, where teams tend to play a softer non conference slate before they run the conference gauntlet.