Experience and Pac-10 wins

Despite the average, every Pac-10 team doesn't win precisely two games more or less than the year before. Perhaps the single factor that best predicts whether a team will rise or fall from one year to the next is the experience that squad returns. We use seven years of Pac-10 data to predict what returning experience suggests about the 2009 Pac-10 in this "Inside the Numbers".

In our last "Inside the Numbers," we discovered that the average Pac-10 team wins 6.8 games which, on average, will be exactly two victories different than the previous season's win total. That's all well and good (and for Stanford fans, mighty exciting when you let yourself daydream of a future in the Pac-10's upper echelon, given today's recruiting classes), but us Cardinalmaniacs know all too well that, despite the average, every Pac-10 team doesn't win seven games each season. Using some additional data to more precisely predict this upcoming Pac-10 season would be lovely, no?

Luckily, we also looked at each squad's returning experience, recording the number of total returning starters, returning offensive starters, returning defensive starters and whether the starting quarterback returned for each season in our database: the 70 total seasons (seven apiece) Pac-10 squads have played since 2002. We can then attempt to correlate how much of an effect, if any, experience has on a team's improvement from one season to the next. We've alternatively challenged and upheld conventional wisdom in past "Inside the Numbers," and it will be interesting to see whether it is true that experience at the quarterback position is of the utmost importance, or whether, in this day of more and more five-star freshmen seeing the field the first day they step on campus, experience at all matters as much as we suspect it does.

I ran four separate linear regressions, which respectively looked at the effect total returning starters, returning offensive starters, returning defensive starters and a returning starting quarterback had on a Pac-10 team's change in win total across our 60 observations. (Note that, though we've recorded data on 70 seasons, we only have 60 differences in win totals between seasons. I use change in wins instead of total wins to control for the fact that USC has always won more games than Arizona, regardless of how many starters each team returned.) A priori, I would expect returning offensive experience to matter most (memorizing hundreds of specific plays has to have a steeper learning curve than memorizing your team's defensive packages, right?), with returning quarterback experience mattering most of all. Turns out I was half-right.

Expected win difference = 0.28*returning starters – 3.7
Expected win difference = 0.34*returning defensive starters – 2.2
Expected win difference = 0.22*returning offensive starters – 1.6
Expected win difference = 1.7*returning QB – 1.3

Note: While the correlations between all these factors certainly doesn't approach the level of statistical significance, they are reasonably consistent between teams, between seasons and with common sense expectation, all of which should give us confidence that we're onto something here. Email me at Daniel at thebootleg.com if you'd like the raw data and associated graphs in Excel.

Okay, translating into English, Stanford returns 17 total starters, nine offensive starters, eight defensive starters and one quarterback according to Phil Steele, the definitive preseason college football Bible, and the source of all this data. (Listed as returning offensive starters are Tavita Pritchard, Toby Gerhart, Owen Marecic, Doug Baldwin, Ryan Whalen, Jim Dray, Chase Beeler, Andrew Phillips and Chris Marinelli. Listed as returning defensive starters are Matt Masifilo, Ekom Udofia, Erik Lorig, Chike Amajoyi, Clinton Snyder, Kris Evans, Austin Yancy and Bo McNally. A player counts as a returning starter if he's started at least six games, or approximately half a season, over the course of his career.)

We can plug Stanford's numbers into the formulas above to see what seven years of Pac-10 history suggests about how Stanford's 2009 record will deviate from its 5-7 2008 mark. Given all the experience Stanford returns, this should be a fun exercise.

Looking at all returning starters, Stanford should improve 1.1 wins, to a 6-6 mark. Looking at returning defensive starters, Stanford should improve 0.9 wins, again to a 6-6 mark. Looking at returning offensive starters, Stanford should improve only 0.2 wins, staying 5-7. Finally, looking at a returning quarterback, Stanford should improve 0.4 wins, I guess to 5-6-1 or something, but, functionally, if the Card start redshirt frosh Andrew Luck, you'd probably expect the Card to slide to 4-8. (Because each of these models was derived independently, we'd be double and triple-counting if we added up the expected improvement for a Pritchard-led Card to generate a 9-3 final mark. Though that would be nice.)

Note that this accounting looks at the effect of experience alone. If you believe Stanford's overall talent level will be better than last year or they'll be more familiar with the coach, or the two additional home games will have an effect, then all of that warrants additional adjustment to your expectations.

All teams, however, are not created equal, and so we can further splice the data to see what effect returning experience has on each Pac-10 program. For USC, for example, we'd expect returning experience to have virtually no effect, since the Trojans win 11 or 12 games every year regardless of who comes back, and even when a starter does leave, his replacement is often All-Conference in short order. Indeed, the numbers cancel out perfectly: returning experience has had exactly zero effect on any change in USC's record. So not to channel my inner Matt Squeri, but rumors of USC's demise may prove greatly exaggerated, even though the Trojans return just three defensive starters this year.

On the other hand, the experience factor plays exactly to form for the Oregon and the Washington schools, programs traditionally in the middle of the Pac-10. For these four teams, an increase in the number of returning starters is strongly correlated with an improvement in record come next year. I think the rationale is that unlike USC, those four schools are simply not good enough where they can maintain their program at the same level regardless of attrition. Indeed, two of the biggest collapses in recent Pac-10 memory came when these schools had few returning starters. Washington returned only 11 starters (14 is about average) and no quarterback in 2004, and they dropped from a 6-6 to a 1-10 mark that season. Washington State returned just six! starters the same season and dropped from 10-3 to 5-6. Oregon State, on the other hand, returned 16 starters and a quarterback in 2006, when they improved from 5-6 to 10-4. The Beavers had lost at least five games each of the five previous seasons; they have lost just four games in each of the three seasons since.

For UCLA, Arizona State and Cal, meanwhile, returning experience has had little discernable effect on final record, which is hardly unexpected given that these three teams have been the Pac-10's most mercurial and enigmatic since 2002. Most inexplicably, UCLA returned 20 starters off a 7-6 2006 squad that finished its regular season with three straight upset victories, the last a 13-9 shocker of then-No. 2 USC. They were 2-0 and No. 11 nationally two weeks into 2007, then dropped a 44-6 decision at unranked Utah and, for good measure, five of their last six to finish 6-7.

(For what it's worth, Stanford's record has fluctuated plenty over these past seven seasons, but it has returned 14 or 15 starters for each of the past five seasons, preventing us from drawing any conclusions on the specific effect returning talent has for the Card.)

All this for the grand finale:

Team, last year's record, returning starters, experience's effect for this program
USC 12-1,12, none
Oregon, 10-3, 9, lots
Oregon State, 9-4, 10, lots
Cal, 9-4, 15, none
Arizona 8-5, 13, little
Arizona State, 5-7, 13, none
Stanford, 5-7, 17, unknown
UCLA, 4-8, 16, none
Washington State, 2-11, 15, lots
Washington, 0-12, 18, lots

Again, schedule strength, overall talent level and countless other factors are not included, but here is what the returning experience alone suggests:

Down arrow: Oregon, Oregon State
Up arrow: Washington, Stanford?

Four teams, two different projected directions. It'll be interesting to see if these squads' seasons play to form.

Are you fully subscribed to The Bootleg? If not, then you are missing out on all the top Cardinal coverage we provide daily on our award-winning website. Sign up today for the biggest and best in Stanford sports coverage with TheBootleg.com (sign-up)!

The Bootleg Top Stories