Writers of Pro Football Prospectus 2008

Most Recent FO Features


» 2018 Free Agency Cost-Benefit Analysis

Is Kirk Cousins the best free-agent quarterback in recent memory? Should Trumaine Johnson or Malcolm Butler have gotten the larger contract? And what makes a free-agent contract good or bad, anyway?

10 Sep 2008

FEI Week 2 Ratings

by Brian Fremeau

The Fremeau Efficiency Index (FEI) principles and methodology can be found here. Like DVOA, FEI rewards playing well against good teams, win or lose, and punishes losing to poor teams more harshly than it rewards defeating poor teams. Unlike DVOA, it is drive-based, not play-by-play based, and it is specifically engineered to measure the college game.

FEI is the opponent-adjusted value of Game Efficiency (GE), a measurement of the success rate of a team scoring and preventing opponent scoring throughout the non-garbage-time possessions of a game. Like DVOA, it represents a team's efficiency value over average.

Only games between FBS teams are considered. Since limited data is available at the beginning of the season, the ratings to date are a function of both actual games played and projected outcomes based on the 2008 Projected FEI Ratings. The weight given to projected outcomes will be reduced each week until mid-October, at which point the projections will be eliminated entirely.

Rank Team Rec FEI
1 USC 1-0 0.265
2 Ohio State 1-0 0.264
3 Georgia 1-0 0.262
4 Florida 2-0 0.234
5 Oklahoma 1-0 0.216
6 LSU 0-0 0.212
7 Missouri 1-0 0.185
8 Auburn 2-0 0.185
9 Virginia Tech 0-1 0.181
10 California 2-0 0.177
11 Wisconsin 2-0 0.172
12 Florida State 0-0 0.167
Rank Team Rec FEI
13 BYU 1-0 0.160
14 Alabama 2-0 0.155
15 Oregon 2-0 0.145
16 Texas 2-0 0.140
17 Texas Tech 1-0 0.136
18 Utah 2-0 0.128
19 Michigan 1-1 0.119
20 Notre Dame 1-0 0.116
21 Penn State 1-0 0.114
22 Michigan State 1-1 0.111
23 Tennessee 0-1 0.110
24 Oklahoma State 2-0 0.110
25 East Carolina 2-0 0.110

The complete Week 2 Ratings for all 120 FBS teams can be found here.

It is still too early in the season to make absolutely conclusive statements about teams. But even though the projected results still carry significant weight, a couple of things stand out in the ratings. A total of three non-BCS conference teams appear in the FEI top 25 (BYU, Utah, and East Carolina), one more than the combined total of the ACC (two) and Big East (zero). Virginia Tech will drift down in the coming weeks if they don't meet their projected expectations, and the Hokies are only hanging in the top 10 now on the strength of those projections and a close loss to what looks to be a very good East Carolina team. No. 12 Florida State won't face its first FBS opponent for ten more days, so it looks like they're safe for now.

I still expect several teams in these "power" conferences to recover some dignity from early season letdowns, but the cumulative effect of multiple-score losses to South Carolina (by North Carolina State), Alabama (Clemson), Bowling Green (Pittsburgh), Northwestern and Akron (Syracuse), USC (Virginia), Fresno State (Rutgers), East Carolina (West Virginia), Florida (Miami), Middle Tennessee (Maryland) and Oklahoma (Cincinnati) will never be fully overcome. The ACC and Big East in total have recorded a grand total of two wins against out-of-conference BCS competition, including a 41-13 victory by Wake Forest over lowly Baylor that barely qualifies. Forget the impending BCS controversy if both conference champions finish behind multiple non-BCS league teams. I want to know if these conferences will be able to qualify 14 teams to fill out their bowl bids at the end of the season.

Home Field Advantage

When FEI was first introduced two years ago, the column included a single paragraph on Home Field Advantage, largely dismissing it as statistically insignificant. Believe it or not, this frustrated me as much as it frustrated some FO readers. HFA is supposed to be one of the eternal truths in sports, perhaps even in life. Death, Taxes, Home Field Advantage. Playoff systems in sports other than college football are deliberately structured to reward high-performing teams with the advantages of playing at home. Fans create hostile environments for visiting opponents and motivate the home team with feverish support. The negative effects of travel and foul weather, the potential for unfavorable officiating, and the unfamiliarity of the other guys' turf are all supposed to negatively impact the visitors. So why did FEI dismiss it then, and better yet, what's happening going forward?

The question for me was never about whether HFA could or could not be observed. The question for me was how precisely it could be measured and how it could be used. Commonly, HFA is understood to be worth three points per game in the NFL, and somewhere between three and five points in college football. Three is a quick-and-dirty number that's easily agreeable for both the casual and educated fans, but what is it supposed to mean in practical terms? Are teams, on average, supposed to make an extra field goal at home that they would have missed at a neutral site? Is it more accurate to name it Road Field Disadvantage, and punish the visitor rather than reward the host team? Does that distinction really matter? And biggest question for me: Does mis-matched opposition skew measurements of how much college football HFA is really worth? Can we calibrate it somehow?

Two weeks ago, I introduced Projected Win Expectation as a method to project college football team records from FEI data. As it turns out, there is a very strong relationship between PWE and Power Advantage, the difference in standard deviation of two teams' FEI ratings. Figure 1 plots the PWE/PA relationship for all teams in blue and home teams in red.

Projected Win Expectation vs. FEI Power Advantage

Teams clearly increase their likelihood of winning by playing at home, particularly in games between teams of relatively equal power. Games that would be 50/50 on a neutral field are won 60 percent of the time by the home team. PWE is boosted by at least five percent for home teams in more than two-thirds of all college football games every year. But as the power advantage grows, of course, the HFA effect on PWE becomes negligible and irrelevant. A team that is likely to dominate its overmatched foe and win 98 percent of the time on a neutral field can't elevate that likelihood much more by playing at home.

Figure 2 represents the full Home Team PWE Advantage distribution for both team Power Advantages and Disadvantages. Note the imbalance in the graph around the zero PA axis. With the exception of significant mismatches, playing at home is markedly more valuable to teams with a Power Disadvantage than it is to home teams with a reciprocal Power Advantage. In other words, if the No. 10 and No. 25 teams play, the PWE Advantage would be greater if the game were at the lower ranked team's stadium than if it were at the higher one. Home underdogs are popular picks for upset specials, and the distribution in Figure 2 helps illustrate why.

Home Team PWE Advantage vs. FEI Power Advantage

Interestingly, the effect of HFA on Game Efficiency, the foundation of FEI, is not at all dependent on Power Advantage or Disadvantage. Home teams play on average about four percent more efficiently than their road opposition, and that average holds up for large PA games as well as evenly-matched ones. Converting GE into points, however, is dependent on another variable: the number of possessions in the game. Solving the Game Efficiency formula for a national average number of game possessions, 24, and HFA in terms of points calculates to 3.4 points per game. As the number of competitive possessions in the game increases or decreases, the value of HFA proportionately follows suit at just under 0.5 points per three possessions.

Pace and style of play are big factors in determining the number of competitive possessions in a given game, but teams don't categorically play a constant possession-per-game rate. Nor do they fit a certain style. Florida and Navy averaged the fewest competitive possessions per game last season (21 per game), and both had very efficient, though very different offenses. Navy's ball control running attack generally reduced overall game possessions; Florida frequently raced out to fast leads, triggering garbage time earlier in the game.

In summary, Home Field Advantage is clearly measurable, but it can be calculated and used by FEI in different ways. Beginning with this week's ratings, the weekly FEI adjusts Game Efficiency data to compensate for the four percent home team GE boost and the weekly FEI Forecasts for all games published here include Projected Win Expectations adjusted according to the Figure 2 distribution. Home Field Advantage is still somewhat intangible, and further research could illuminate other measurements of team-specific HFA, weather-specific HFA, and more. Any and all suggestions are welcome.

Posted by: Brian Fremeau on 10 Sep 2008

16 comments, Last at 12 Sep 2008, 9:58am by Ross


by joe football (not verified) :: Wed, 09/10/2008 - 6:55pm

Well, ND being #20 is a little silly, given they're coming off a horrible year and struggled to win at home against a team that lost to a non-FBS team the week prior and was ranked 108th in pre-season FEI. ND was ranked 28th in pre-season FEI, which I assume is due to their Program FEI being 25th. However, they were 33-28 over the 5 year span considered by Program FEI and typically have an average at best schedule outside of one or two games, so I have to question that as well.

In conclusion, FEI is intentionally biased towards Notre Dame and should be scrapped in it's entirety

by Peter Libero (not verified) :: Wed, 09/10/2008 - 7:24pm

Notre Dame is clearly ranked too high because of its historical tradition and media bias. Ranking entirely based on 2 weeks is way better than this. BOO NOTRE DAME, GEORGETOWN = REAL JESUITS!

by Jeff (not verified) :: Wed, 09/10/2008 - 8:18pm

these rankings are horrible.
ND at #20? Please.
BYU, who beat UW by 1 (thanks to the refs) ahead of Oregon who beat them by 34 (with 3rd and 4th string QB)... really? come on...

by Will B. (not verified) :: Wed, 09/10/2008 - 8:47pm

I hate early rankings - despite all the disclaimers that people put in to them, ESPECIALLY in computer rankings, people still rush and call them dumb. By the end of the season, FEI will be a very legitimate ranking. Early in the year, it's just guessing based on previous seasons.

"Since limited data is available at the beginning of the season, the ratings to date are a function of both actual games played and projected outcomes based on the 2008 Projected FEI Ratings. The weight given to projected outcomes will be reduced each week until mid-October, at which point the projections will be eliminated entirely."

by mm (not verified) :: Wed, 09/10/2008 - 8:54pm

I would be interested to see if you could further separate home field advantages into more components:

1)Turf advantage: teams try and build there teams around their stadium. A team built for an artificial surface may have a bigger HFA hosting a team built for a grass surface and vice-versa. However, they may host a team that is even better suited for the artificial surface than they are, causing them to 'lose' this aspect of HFA.

2)Weather advantage: some teams are also built to handle inclement weather, while others depend on precision and footwork that may be disrupted by bad weather. Again, it might be possible to 'lose' this aspect of HFA as well, to a teams that's actually better designed to handle the conditions (whether those conditions are good or bad). This would be much harder to measure than the turf advantage, because the weather conditions can change during the game, and the write up in the box score might not accurately represent the field conditions.

3) Cheering advantage (from fans in the stands): This one might be possible to measure. Does this aspect of HFA rise when more fans are in the stands? If Directional U hosts the big University from its own state and finds the stands half-filled with fans of the opponent, does HFA drop?

4) HFA from familiarity. Sleeping in your own bed, and going to a locker room and sidelines that you're used to may provide a boost. I think you can only really measure this by trying to measure each of the above aspects of HFA and then subtracting that from the 'average' HFA value.

Can anyone think of other possible parts of HFA?

by bradluen (not verified) :: Wed, 09/10/2008 - 10:02pm

OK, one more time... BYU outplayed Washington, the Huskies couldn't do anything to stop BYU's offense, the game wouldn't have been close if BYU had recovered a third quarter fumble on Washington's 1, BYU had a better than even chance of winning if it went to OT.

Notre Dame at #20, I'm not going to try to defend that.

by Ken Shabby (not verified) :: Wed, 09/10/2008 - 10:27pm

Come on, people - Notre Dame got a healthy bump after beating perennial powerhouse San Diego State

by Pat (not verified) :: Thu, 09/11/2008 - 12:14am

Yeah, Notre Dame is definitely bizarre. Especially because the team below them, Penn State, was projected higher in the FEI Projections (0.151 vs 0.119). Notre Dame and Penn State have each played 1 FBS team: Notre Dame played SDSU (projected at -0.151, #102) and Penn State played Oregon State (projected at 0.091, #35).

Penn State demolished OSU (45-14), and Notre Dame barely eked by SDSU (21-13).

Yet Notre Dame still ends up above Penn State? There's got to be a mistake (as in, literal mistake - I don't think Brian's biased, I just think there's a bug somewhere) - Penn State has a higher Projected FEI Rating, and has to have a higher FEI for their game (assuming 3 quarters of 'competitive drives' for Penn State, they had 45 points on 10 drives, and allowed 14 points on 11 drives - assuming all 4Q of 'competitive drives', Notre Dame had 21 points on 15 drives, and allowed 17 points on 15 drives), and their opponent in the game has a higher Projected FEI Rating (and a higher FEI Rating).

As far as I can tell, all the components that go into the current week FEI rating should be higher for Penn State than Notre Dame, and not by a little, but by a lot. So something's definitely wrong.

by bradluen (not verified) :: Thu, 09/11/2008 - 1:13am

If there's a bug, it's more likely in Penn State's rating than Notre Dame's. Looking at the numbers on Brian's blog, Penn State's week 1 rating (after beating Hapless FCS Team 66-10) was a lot lower than their preseason rating.

by ammek (not verified) :: Thu, 09/11/2008 - 4:30am

5: Other factors include: home fans putting pressure on the officials; and testosterone levels (see link).

by GoodKingJohn (not verified) :: Thu, 09/11/2008 - 8:23am

10 and 5.. interesting to break down into components. cant think of any others. however, testosterone level, I would think is a by product of number 3. of the five mentioned number 5 being pressure on officials), how you rank in order of importance and give weightings (educated guess) that sum to 1?

3 (.6),1 (.17),2 (.13), 4(.06), and 5 (.04)

by Will B. (not verified) :: Thu, 09/11/2008 - 12:44pm

RE: HFA - I bet distance travelled has to play in to it a bit as well (although this would more accurately be captured as road team disadvantage). Ohio State, for instance, has to travel a long way and play much later than they typically would have to.

by Pat (not verified) :: Thu, 09/11/2008 - 1:51pm

#9: Got me which team it is. Notre Dame had a projected FEI rating of 0.119, and then in Week 1 (after not playing anyone!) had an FEI rating of 0.141. So something's just weird, in general.

by ammek (not verified) :: Fri, 09/12/2008 - 5:19am

"testosterone level, I would think is a by product of number 3 (cheering)"

Not necessarily. Although it's a small sample size, research has suggested that soccer teams playing at home in front of an empty stadium (usually as a punishment for crowd trouble) retain a (smaller) HFA. Equally, teams like Scotland's Queens Park show no reduction in HFA, even though it attracts no more than 1,500 spectators in the national stadium with a capacity that has ranged from 50,000 to 90,000. Crowd noise can't be much of a factor there.

Similarly, as the article I linked noted, the researchers found increased levels of testosterone in players several hours before the game began - suggesting that there's some neuro-/psychological mentality at work among home players. This may well be linked to the desire to avoid shame in front of supporters, or to the weight of expectations.

by Ross (not verified) :: Fri, 09/12/2008 - 9:54am

Notre Dame being ranked above Penn State in ANY ranking system makes me seriously question the rankings.

by Ross (not verified) :: Fri, 09/12/2008 - 9:58am


PSU's week 1 win means NOTHING - only games against other FBS teams are considered.