September 29, 2010

Utah by the Numbers . . . .

Today, I take a look at Utah in a few statistical categories after four games into the season.

First off, Utah's Offensive line has been phenomenal. I'll be the first to admit that their first few games haven't been the toughest. However, sacks are given up even to the worst of teams, so the fact that Utah has only given up a single sack through four games speaks volumes of the Offensive line unit (stat). Additionally, you'll remember that Utah's first opponent (Pittsburgh) boasted two of the nation's premier pass-rushing DE (Romeus, Sheard). Pittsburgh led the nation in sacks the previous year (stat), and returned their two top sack leaders from '09. For Utah to have only given up 1 sack thus far is impressive.

Additionally, Utah has only given up 6 TFL in 4 games, a number which leaves them tops in the country through 4 games (stat). This stat also indicates Utah's Line is not allowing penetration into the backfield and keeping gaps open long enough for positive yardage gains on every play. This could dip as the schedule gets tougher, but to be sitting on top after 4 games is a very positive sign, regardless of who Utah has played.

Tackles for Loss Allowed

2Penn St.464478.02.00

Sacks Allowed

1Middle Tenn.41.05.25
1San Diego St.41.06.25
1Penn St.41.019.25

Predictive Statistics (Pass Efficiency Differential Margin)

One statistic which I enjoy measuring for predicting outcomes is the pass efficiency differential margin (PEDM). This is obtained by taking the difference between the offensive and defensive pass efficiency ratings for a team.

Pass efficiency (PE) is a measure of a team's passing ability, which is measured by four categories: (1) yards per pass attempt, (2) pass completions per pass attempt, (3) touchdowns per pass attempt and (4) interceptions per pass attempt. In the NCAA formula, four constants (i.e., 8.4, 100, 330, and 200) are used such that an average passer will have a rating close to 100. Pass efficiency is explained in more detail (here). Pass Efficiency Defense is calculated using the same formula, except it measures the PE of the opposing quarterback from week to week. Cumulatively, as opposing QB numbers are tallied each week, the Pass Efficiency Defense is determined from the cumulative opposing QB numbers.

The Pass Efficiency Differential Margin (PEDM) is the difference between Pass Efficiency (PE) and Pass Efficiency Defense (PED), and I believe it is a good predictor for wins and losses.

There are of course many other indicators that can determine the outcome of a football game, such as home/away, turnover margin, weather, special teams play, etc., etc. However, I believe that this metric (PEDM) is particularly suited to predict Utah football outcomes for one reason: Utah's Defense has consistently shown it can stop the run over the years. Their defensive system hasn't changed for years under Kyle Whittingham, and it has long been known for imposing defensive lines, filling up the box, and stuffing the run. In other words, I'm not paying much attention to the run defense here, since I'm assuming this is generally a constant under Utah's defensive scheme, and they are not often going to get beat by giving up massive amounts of rushing yardage.

With that disclaimer stated, let's look at the numbers: DATA

Since I'm using the difference between team's PEDM to predict wins and losses each week, I use the cumulative data from the beginning of the season up to the week before the game to be predicted. For example, to predict the Iowa State vs. Utah game winner, I'd include all the data for each team prior to that game.

A few things should be pointed out when using this analysis:

(1) This predictive power of this metric becomes much more reliable as the season wears on. Early on, teams with hard / easy schedules may not reflect an accurate PEDM.
(2) As the conference play approaches and teams begin to play each other within the same conference, the metric becomes much more meaningful.
(3) Obviously, the PEDM metric can be effected suddenly when QB are injured, or switched. For example, a team with a great QB1, but a terrible QB2 might have a great rating after 5 games. But if QB1 gets injured, the metric might easily predict a win, even though QB2 is going to hurt those chances significantly.
(4) As with (3) above, this is another reason why I feel this metric is particularly suited to predict Utah outcomes this year, since we've already seen good levels of production from QB1 (Wynn) and QB2 (Cain).

A quick look at the data reveals some interesting observations:

(1) Pass efficiency differential margins (PEDM) are shown in the colored column for Utah's opponents (left) and Utah (right). The Legend at right shows what the colors mean. Green means you're a good team . . . and red means . . . you're not. The greener the better, the redder . . . the worse. You get it.

(2) Since no data is available for the first game of the season (PITT), I used the PEDM for both teams from the conclusion of 2009 (post bowl-game). A positive difference of 1.97 is a small positive margin, indicating a narrow win. Utah beat Pitt 27-24 . . . in OT.

(3) UNLV and NM games show PEDM differences of +79.27 and +110.89 in Utah's favor, easily predicting wins, and they were . . . blowouts. The game against SJSU shows a PEDM diff of +91.26 . . . another blowout.

(4) The ISU game currently shows a +72.22 PEDM difference in Utah's favor, but this number will change after ISU plays its week 5 game against Texas Tech. I'll update the number then, but it's not going to change enough to effect the prediction of a W! The number indicates that this should be the toughest game since Pittsburgh, but just slightly harder than UNLV. Given that the game is on the road, it will be a bit tougher. But this should still be a big win for Utah.

(5) I've shown the PEDM data for each of Utah's opponents up to the current week. I will only update the information for the teams Utah hasn't played, so that the PEDM data reflects what it was the week before the game was played. For example, this week I'll update ISU's data and all teams further out, but will not update any teams data prior to ISU.

blog comments powered by Disqus