Interview With College Football Outsiders Statistician Brian Fremeau
The stat gurus at Football Outsiders have released the most recent edition of their off-season annual—Football Outsiders Almanac 2010—complete with an introductory explanation of some new and innovative statistics, 2010 projections for every college football team in the Football Bowl Subdivision, quarterback and recruiting rankings, and a brief history of conference realignment.
In conjunction with this release, Clashmore Mike sat down for a Q&A with almanac author Brian Fremeau. Among the topics discussed: improvements in this year’s almanac, potential breakout teams in 2010, and the critical pieces needed for a successful season in head coach Brian Kelly’s first year.
Brian has been published at ESPN.com, FootballOutsiders.com, and in the Maple Street Press college football publications—including this year’s edition of Here Come The Irish where he details the tempo of Kelly’s spread offense. He is also the creator of the Fremeau Efficiency Index (FEI), a drive-based efficiency metric seven years in the making.
Introduction in hand, let’s proceed to the questions…
This is the third year of the College Football Outsiders Almanac and the second in which you’ve contributed. What was the impetus behind it, how has it evolved, and what is new and improved in the 2010 edition?
We have been developing ratings and college football statistics at Football Outsiders for several years. Most of that work has been focused on retroactive analysis to better understand and evaluate team success in the rearview mirror. When we discovered that there was a stronger year-to-year correlation of our opponent-adjusted data than we had originally hypothesized, we developed a projection model for the annual almanac.
The basis for the projection model is a five-year program success rating, adjusted with transition factors for the upcoming season. We collected and tested more data than ever before to create this year’s projection model—returning starters, recruiting ratings, talent lost to the draft, unsustainable success (e.g. high fumble recovery or outlier production on passing downs), and more.
In addition to making the projection model more robust, we’re debuting new “win likelihood” tables that detail the probability of any given record for every team. We think these tables demonstrate the relationship between projected team strength, schedule strength and record in a more comprehensive way.
College Football Outsiders specializes in non-traditional, but incredibly insightful, college football statistics and metrics. What are some of the most recent developments and improvements and how do they differ from previous versions?
The most noteworthy new development is our F/+ metric, a combination of two entirely separate rating systems. My system (FEI) is based on drive analysis, the success of a team maximizing its own possessions and minimizing those of its opponent. My colleague, Bill Connelly, works with play-by-play data, combining a measure of efficiency and explosiveness in his S&P+ rating system.
Both systems collect raw data and adjust for opponent strength. Both systems discount garbage time plays and possessions. But both systems are independently produced, make opponent adjustments in unique ways, and sometimes produce very different results. We created the F/+ metric as a way to arrive at a consensus between our two approaches, and discovered this off-season that F/+ had a stronger correlation to next year success than either of our independent rating systems. We will continue to produce FEI and S&P+ ratings and analysis this fall, but we’ll be featuring F/+ data in a more prominent way during the season as well.
Based on the 2010 predictions detailed in the almanac, what teams are poised to have breakout seasons and why?
Interesting question. To be perfectly honest, our system doesn’t typically identify any “breakout” teams. Since five-year success is the starting point, F/+ isn’t inclined to identify too many programs coming out of nowhere. Instead, we’re likely to project more modest expectations for other teams’ supposed “breakout” hype. Nebraska is an example of a team receiving plenty of top-10 preseason attention, and our system thinks they’ll be good, but not great, and certainly not elite. The Cornhuskers aren’t likely to be quite as strong defensively this year without Suh, and their offense was a liability for most of 2009.
Some of our projections are surprising, certainly. Despite major personnel losses, we think Alabama and Florida will still control the SEC and remain national title contenders. We think TCU is a better pick for a non-AQ run at the national championship game than Boise State. Tennessee is the biggest surprise in our projected top-25 (unfortunately, we don’t have a metric to account for bar brawl stupidity), though due to a brutal schedule, we are more conservative with our projected record for the Volunteers than their ranking might imply.
On to the Irish-specific questions. What two or three opponents project to give Notre Dame the toughest test in 2010 and why?
Our projection model forecasts three teams on Notre Dame’s schedule with a higher preseason F/+ rating than the Irish: USC, Pittsburgh and Boston College. But that doesn’t mean we expect the Irish to lose those games and win the other nine. Instead, we produce game-by-game win probabilities for the whole schedule, and those are used to project our overall win probabilities for the season.
USC and Boston College stand out as the toughest tests for Notre Dame according to our model. For starters, both are road games, which can account for up to a 20 percent difference in win likelihood per game than if the games were played at home. We expect both the Trojans and Eagles to boast top-20 defenses that ought to be tough against the run, an area that remains a question mark in Notre Dame’s offensive future under Kelly. We anticipate near toss-ups against Michigan State and Pittsburgh; the Panthers are our top-rated Big East team and figure to have another strong running game that can possibly control and disrupt the Irish tempo, and the Spartans return a decent core from last year.
Notre Dame is projected to be favored against the rest, but is likely to win only three of four against Utah, Stanford, Navy and Michigan. It all adds up to a “mean wins” projection of 8-4.
What will potentially limit success and what assets do the Irish have that are critical components to winning those eight games? Are there any fundamental problems (e.g. tackling on defense) that can be easily corrected by a new coaching staff and quickly produce a positive change in performance?
Offensively, the Irish appear poised to keep rolling—if not quite at the level of last season, then close to it. Obviously Kelly’s track record at Cincinnati suggests he is capable creating an offensive juggernaut, and there are athletes all over the field that can fill the void left by quarterback Jimmy Clausen and wide receiver Golden Tate. One red flag from last year’s Bearcats team that I’ll be watching closely with Notre Dame was their boom-or-bust tendency—their three-and-out rate was abnormally high for an elite offense. If Dayne Crist can move the chains on that first series of nearly every drive, they’ll be in great shape.
The defense was so out of sync last season, its hard not to expect at least some natural improvement this year. The recruited talent is maturing, and Kelly appears to have the entire staff on the same page, unlike his predecessor. Our college data doesn’t identify specifics like tackling fundamentals and their impact on defensive ratings, but there’s plenty of anecdotal evidence to suggest that a new staff with a renewed focus on such details can right that ship. But actually, we’ll be launching a new college football game-charting project at Football Outsiders this fall that will help us collect better data to evaluate more detailed observations of every play.
Clashmore Mike would like to thank Brian for taking the time to entertain our whims. Those interested in picking up a copy of the Football Outsiders Almanac 2010 can go here for all the ordering details. If you wish to purchase only the college portion and study up on the NCAA football odds for all FBS teams, it is available here for a reduced rate.