Football Outsiders will soon release its inaugural College Football Almanac, which should available for purchase today according to FootballOutsiders.com. Their full 2010 almanac is also coming out next week.
Rock M Nation's Bill Connelly is also one of FO's college football writers, so it's only natural that they have a partnership with SB Nation. Owing to my incessant pestering, Bill kindly agreed to answer a few questions today in order to whet everyone's appetites for CFA 2010 and FOA 2010. Interested readers may also want to check back with On the Banks's January Q&A with Brian Fremeau, the other college football outsider. Let's begin.
On the Banks: FO's S&P+ ratings system last year ranked Rutgers as the 78th best team in the country. Is it correct to guess that the primary causes of that ranking were a poor strength of schedule, and benefiting from a large turnover margin?
Bill Connelly: Schedule strength definitely punished Rutgers quite a bit, as it did to Kansas State. The idea behind the strength of schedule adjustment is that if you absolutely crush bad teams or FCS teams, your ratings will not suffer. But when you play two terrible FCS teams, you better beat both of them 65-0, otherwise your ratings will probably go down a bit. Beating them a combined 87-7 appparently wasn't quite good enough.
Aside from that, Rutgers seemed to suffer from a lack of offensive efficiency. Their offensive rank of 101st overall stemmed mostly from their rank of 115th in Success Rates (96th in Rushing Success Rate, 116th in Passing). Inefficiency is to be expected from such a young offense, and inefficiency is what Rutgers had. The good news, of course, is that that will improve significantly with experience.
On the Banks: One of the main principles of Football Outsiders is that there's a significant distinction between forcing and recovering fumbles in the NFL, with the latter not showing a correlation with year-to-year performance. College football has many more teams/games and a wider talent disparity than the pro game. Would it be more informative to look at fumble recoveries in college as opposed to pure turnover margin as a predictive tool? How useful is it to look at a raw total here if many of them come against bad OOC competition?
Bill Connelly: Over the last few years, the correlation between turnover margin and a change in the next year's performance is -0.28. The correlation between the percentage of all fumbles recovered and a change in the next year's performance is -0.15. That means that turnover margin is still a bit of a stronger predictor (too high or too low, and there will almost certainly be a turnaround of sorts the next season), but we took both into account in this year's projections. (Who the turnovers came against would quite possibly be telling, but that was not taken into account in the projections.)
On the Banks: Statistically, are games against FCS teams or low-majors all that informative? Is it more meaningful to beat Eastern Michigan instead of Texas Southern?
Bill Connelly: With such a small sample of games and plays, we have to use all the data we have at our disposal. For the S&P+, FCS teams are broken into six tiers based on their overall level of success. If each "tier" of teams were an FBS team, here is how they would rank in a 126-team FBS (120 FBS teams plus six "tier" teams):
FCS Tier 1: 109th on offense, 97th on defense
FCS Tier 2: 114th on offense, 118th on defense
FCS Tier 3: 115th on offense, 120th on defense
FCS Tier 4: 122nd on offense, 122nd on defense
FCS Tier 5: 123rd on offense, 125th on defense
FCS Tier 6: 126th on offense, 126th on defense
So to answer the question, there really isn't a difference between Eastern Michigan and a FCS team, but we have to squeeze whatever information we can out of every game. And here's where looking at non-garbage time possessions comes into account. Brian (Fremeau, who creates the FEI formula) and I have different ways of determining what is "garbage time," but with the S&P+ formula, it very much behooves you to put games against bad teams out of reach as soon as possible. Rutgers did that pretty quickly with Howard and Texas Southern, but not quite quickly enough, I guess.
On the Banks: Without stepping on anyone's incentive to buy the almanac, is there anything particularly odd or novel regarding the 2010 projection for Rutgers? There's no way without enslaving an army of interns to determine whether or not they really do blitz more than other teams, right?
Bill Connelly: The only way we can approach blitz statistics right now is to compare a team's sack rates on standard downs and passing downs. In theory, a team with better rankings on passing downs than standard downs probably blitzes pretty effectively (since passing downs are where a majority of blitzes are likely to occur). It's not a perfect way of gauging blitzes, but until game charting takes off at the college level (by the way, we're looking for game charters for this coming season! E-mail me at billconnelly1 at gmail dot com if you're interested!), it's probably the best we can do.
I don't believe the almanac goes into detail with sack rates much, but here are Rutgers Adj. Sack Rate ranks for the last five years:
2005: 14th overall (16th on standard downs, 55th on passing downs)
2006: 3rd overall (14th on standard downs, 3rd on passing downs)
2007: 3rd overall (14th on standard downs, 16th on passing downs)
2008: 63rd overall (63rd on standard downs, 24th on passing downs)
2009: 55th overall (29th on standard downs, 15th on passing downs)
In three of five seasons, Rutgers has ranked higher on passing downs, which suggests that maybe a disproportionate amount of their pass rush comes from blitzes.
On the Banks: You mentioned in a recent podcast with In the Bleachers that quarterbacks typically make their biggest jump between their freshman and sophomore seasons. Without giving too much away, how did you go about reaching that conclusion? I ask because very few quarterbacks start as freshmen (even less as true freshmen). When I sort of looked at true freshmen quarterbacks last year (note: link isn't all that informative, it was a victim of severe time constraints), there didn't seem to be any hard and fast rules as pertains to first year performance or career progression.
Bill Connelly: One of the essays in the upcoming Football Outsiders Almanac deals with quarterbacks, their level of experience, and their star ratings coming out of high school (using data from 2005-09). There wasn't enough of a sample size to create another sub-category of, say, years of starting experience, so you are correct that the sample size for freshmen and sophomores is smaller than that for juniors and seniors, and that affects the way you'll need to view the data (disclaimers and clarifications are given in the essay). But yes, the basic conclusion is that quarterbacks are more likely to improve by leaps and bounds in their first two years. After that, the level of improvement is consistent, but smaller.
On the Banks: One thing that I presume will be covered in this year's almanac is your research into recruiting rankings, first covered earlier this year in an ESPN Insider article and a Washington Post podcast. Basically, while each specific player ranking of course isn't 100% definitive, collected recruiting rankings have close to as much predictive power as a program's recent win-loss record when it comes to future projections. What were some of the methodological hurdles with your research on this subject?
Bill Connelly: While the only recruiting essay in the book deals specifically with quarterbacks, this is indeed the first year where overall recruiting rankings are taken into account in the projections. On a macro level, those rankings are almost as predictive of future success as recent on-field performance, which is interesting.
Honestly, the biggest leap dealt with rosters and estimations. Just because a player signed with a school and enrolled there for one season does not mean that they will still be on the team 2-3 years after they signed. Rivals.com does a good job in providing a set of recruiting rankings using only players who actually enrolled, but that still doesn't clear up all hurdles that come with using a five-year weighted average (which is what we do in the projections). Eventually I would love to get to a point where we are using the star ratings of players on a team's projected two-deep instead of just overall recruiting points, but that's where we are right now. Despite the inconsistency, the ratings are still quite strongly correlated to future success, which means we will continue to pursue ways to fine-tune how we use the ratings.
On the Banks: Last year's FOA 2009 had far more college football content than previous Pro Football Prospectus editions. What's in store for the new 2010 almanac? Likewise, Football Outsiders also just launched a new college page on their site. Can you give any hints as to anything new that's in the works for this fall?
Bill Connelly: We are still trying to figure out our weekly schedule during the fall, but honestly, there probably won't be a ton of change. The benefit to the FO college page is simply that you will be able to find the content more easily and revisit it to participate in comments threads. With so much content on the FO front page, it is easy for a piece to get drowned out after a day or less. This way, our major pieces -- One Foot Inbounds, FEI, Varsity Numbers, Seventh Day Adventure -- will be more prominent for a longer period of time.
One thing you will definitely see this fall is an increase in college-related Extra Point posts (the links on the right sidebar of the FO page). That is another good way for generating discussion and sharing information, and we are already attempting to bring more college topics to that forum. And I personally would like to get back to doing some more podcasts, but we'll need the time to do so, and ... we'll see.
Beyond that, suggestions are welcome. As I said, nothing is set in stone, so if there is a particular feature you would enjoy, we're all ears.
(That would be college DYAR/individual ratings for position players!)
So there you have it. Check out the new College Football Outsiders page, and give some thought into buying the College Football Almanac 2010 and/or Football Outsiders Almanac 2010. I know I will be, and compiling all gripes and complaints here while completely ignoring their probable love of Ray Rice.