Writers of Pro Football Prospectus 2008

Most Recent FO Features


» Defense and Rest Time

Do defenses really wear out over the course of a game? Do defenses benefit from long drives that give them more time to rest on the sideline? Guest columnist Ben Baldwin investigates.

07 Oct 2009

Week 5 FEI Ratings

by Brian Fremeau

In the ACC chapter of Football Outsiders Almanac 2009, I openly wondered if Miami head coach Randy Shannon could weather the storm through the early part of Hurricanes' schedule. FEI projections were kind to Miami (No. 18 nationally) but were even higher on the prospects for Florida State, Georgia Tech, Virginia Tech, and Oklahoma, the gauntlet of opponents the Hurricanes would face over the first month of the season. If they were to lose all four, could they gather themselves and rip off eight straight against weaker competition to finish with a respectable record?

Shannon doesn't have to worry about the team's spirits. Miami recovered from an early deficit against Oklahoma Saturday night and finished off the Sooners with a back-breaking clock-killing possession over the final four minutes of the game. Now at 3-1, the Hurricanes are No. 11 in the major polls and the most battle-tested team in the land.

Our unique strength of schedule (SOS) metric calculates the likelihood that an elite team would go undefeated against each given team's schedule. The Hurricanes rank third nationally in that category with a 0.146 SOS rating -- a 14.6 percent likelihood that an elite (top 5) team would run the table over Miami's entire schedule. The heavy lifting is now in Miami's rear-view mirror. Miami has played the most difficult schedule by far (0.339), twice as challenging as the average schedule played to date by other BCS conference teams (0.679).

A year ago, five teams faced a four-game stretch more difficult than Miami's this season and survived with only one loss (Table 1).

Team 4-Game
Opponents over 4-Game Stretch 4-Game
Texas Tech .231 at Kansas (W), vs. Texas (W), vs. Oklahoma State (W), at Oklahoma (L) 3-1 9-2
Texas .255 vs. Oklahoma (W), vs. Missouri (W), vs. Oklahoma State (W), at Texas Tech (L) 3-1 12-1
Georgia .273 vs. Tennessee (W), vs. Vanderbilt (W), at LSU (W), vs. Florida (L) 3-1 9-3
Alabama .301 at LSU (W), vs. Mississippi State (W), vs. Auburn (W), vs. Florida (L) 3-1 12-2
LSU .313 at Auburn (W), vs. Mississippi State (W), at Florida (L), at South Carolina (W) 3-1 7-5

Georgia, Alabama, and LSU all appear on the list primarily due to the stratospheric rating of Florida last season -- the other three games weren't exceedingly difficult challenges, but Florida's rating affects the entire sequence. Texas and Texas Tech both had stacked mid-to-late-season schedules, and each was bumped from national championship consideration due to a loss at the end of the four-game series.

Miami's situation is unique in two key respects. By knocking out its toughest challenges early, Miami should be positioned to move up the national polls every week as others fall away. Morale isn't a problem, but the Hurricanes may need to battle malaise and focus issues throughout the rest of the season. Will the swagger carry the day when the whole world isn't watching? This weekend presents a classic letdown situation, but the opponent is lowly FCS (formerly Division 1-AA) squad Florida A&M.

Second, each of the difficult stretches in Table 1 were played exclusively against conference competition. Miami has already played what appear to be three of the most challenging ACC teams it will face, but the Hurricanes' victory over Oklahoma may have the most lasting effect for its BCS national championship aspirations. If a healthy Sam Bradford can put the Sooners back on track for a BCS bowl berth of its own, the computer systems (including FEI) will love the Hurricanes even more.

Week 5 FEI Top 25

The principles of the Fremeau Efficiency Index (FEI) 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. Strength of Schedule (SOS) is calculated as the likelihood that an elite team would win every game on the given team's schedule.

Only games between FBS teams are considered in the FEI calculations. 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 2009 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.

ALSO: FEI ratings for all 120 FBS teams are now listed in the stats page section of FootballOutsiders.com. Click here for current ratings, or here for 2008 ratings. Other past seasons will be added soon.

Rank Team FBS
FEI Last
1 Florida 3-0 0.212 1 0.402 3 0.362 71 0.441 15 -0.680 4 0.571 11
2 Texas 4-0 0.202 4 0.386 4 0.345 65 0.297 28 -0.696 2 0.573 9
3 Virginia Tech 4-1 0.198 2 0.177 21 0.216 14 0.446 14 -0.321 27 0.517 42
4 Miami 3-1 0.198 9 0.003 56 0.146 3 0.159 39 0.102 68 0.511 49
5 TCU 3-0 0.197 10 0.233 12 0.395 81 -0.002 56 -0.660 7 0.544 21
6 Alabama 5-0 0.197 7 0.380 5 0.243 27 0.603 8 -0.695 3 0.536 25
7 Auburn 5-0 0.192 5 0.235 10 0.330 60 0.604 7 -0.253 32 0.510 50
8 Iowa 4-0 0.182 3 0.220 16 0.257 38 0.017 54 -0.670 6 0.551 18
9 Boise State 4-0 0.182 8 0.402 2 0.625 115 0.466 13 -0.624 8 0.574 8
10 Oklahoma 1-2 0.181 12 0.198 19 0.128 2 0.303 27 -0.420 19 0.518 40
11 Ohio State 4-1 0.179 13 0.233 11 0.335 63 0.153 40 -0.511 14 0.559 15
12 USC 4-1 0.167 17 0.237 9 0.229 23 0.094 45 -0.757 1 0.524 34
Rank Team FBS
FEI Last
13 South Florida 3-0 0.162 14 0.223 15 0.253 36 0.343 23 -0.566 10 0.500 60
14 West Virginia 2-1 0.159 6 0.081 37 0.243 28 0.353 21 -0.161 42 0.449 101
15 Cincinnati 4-0 0.156 26 0.342 7 0.309 52 1.326 1 -0.270 31 0.485 83
16 Kansas 3-0 0.152 23 0.343 6 0.304 49 0.490 12 -0.513 13 0.590 6
17 Florida State 1-3 0.150 11 0.038 48 0.115 1 0.115 43 0.545 113 0.597 2
18 Arizona 2-1 0.139 27 -0.007 60 0.221 18 -0.189 79 -0.014 59 0.524 32
19 Pittsburgh 3-1 0.137 32 0.152 25 0.317 54 0.290 29 0.283 96 0.593 4
20 LSU 5-0 0.134 36 0.209 18 0.222 19 0.078 51 -0.466 17 0.543 22
21 Oregon 4-1 0.133 16 0.223 14 0.240 26 0.196 37 -0.535 11 0.530 28
22 Clemson 2-3 0.128 15 0.066 41 0.173 6 -0.461 109 -0.449 18 0.557 16
23 Boston College 3-1 0.124 35 0.101 32 0.218 16 0.144 41 -0.502 16 0.484 84
24 Nebraska 3-1 0.122 42 0.430 1 0.270 43 0.711 4 -0.601 9 0.596 3
25 Utah 3-1 0.120 25 0.117 29 0.321 58 -0.024 58 -0.393 21 0.520 38

Posted by: Brian Fremeau on 07 Oct 2009

13 comments, Last at 09 Oct 2009, 4:04pm by Brian Fremeau


by Fourth :: Wed, 10/07/2009 - 6:07pm

Miami would need Texas to lose or the SEC to really beat up on each other in order to reach the BCSCG. But they would also need VT to lose twice in conference, otherwise the Canes won't win their ACC division. At the same time if VT starts racking up losses, that blowout starts to look even worse for Miami. I just don't see them having a real shot of making it this year (though stranger things have happened). Perhaps next.

by UM Tim (not verified) :: Thu, 10/08/2009 - 12:28pm

Why do we need VT to lose 2 conference games for us to make it to the Championship game? I hope VT does win out but then loses in the ACC Championship game. Our 11-1 squad will be ranked higher than the ACC Champ (whoever that may be).

by BooksNBiscuits (not verified) :: Thu, 10/08/2009 - 5:46pm

We need VT to most likely lose twice because we lost head to head.

The only scenario with us making the ACCCG and VT losing only once is if that loss happens to be to GT, while we run the table. Then, as I understand it, we'll be the Coastal rep in a three-way tiebreaker.

Of course, VT could lose three, too. But I don't see that happening.

On a poll-wide note. FSU that high seems odd. Sure, their losses are to teams with only one loss, but still. They're 2-3.

by Kibbles :: Wed, 10/07/2009 - 6:42pm

I'm loving the new charts.

by Corey (not verified) :: Wed, 10/07/2009 - 11:03pm

Can someone explain, based upon the data presented, why Pitt is higher than Nebraska? I asked this question last week and received no response.

by Kibbles :: Wed, 10/07/2009 - 11:35pm

If I had to guess, it probably has a lot to do with Nebraska having faced FAU, Arkansas State, and Louisiana-Lafayette. Game Efficiency is not adjusted for opponent, while FEI is basically just opponent adjusted GE. At least, that's my understanding of it. Nebraska's GE is off the charts, but after adjusting for schedule, it falls back to earth.

I know that the table says Nebraska has had a tougher SoS than Pitt so far, but FEI's SoS rating is different than most SoS ratings- FEI's SoS is a measure of the percent chance that a top-5 team would go undefeated against that schedule. As a result, a team that had played, say, Florida and a trio of 1-AA teams would have a very tough SoS (because the odds of a top-5 team beating Florida are low), while a team that had played four teams ranked between 40-60 would probably have a relatively low SoS (because a top 5 team should have no trouble dispatching teams in that range). Still, the first team would face a much harsher adjustment from GE to FEI because 75% of its plays were run against 1-AA competition (whereas the second team ran 100% of its plays against middling 1-A competition, and likely wouldn't face as severe of an adjustment).

Just a guess.

by Corey (not verified) :: Thu, 10/08/2009 - 7:46pm

The SOS in the FEI ratings table isn't the FEI SOS? How is that the least bit logical? In fact, it's so illogical I think your premise is incorrect (I'm betting the SOS listed IS the FEI SOS).

BTW, Pitt has played Youngstown State, Buffalo, Navy, NC State and Louisville so far. I don't see how that schedule is much different than Nebraska's. Nebraska's home schedule has been junk, but so has Pitt's.

by ejruiz (not verified) :: Thu, 10/08/2009 - 8:34am

How in the world can you justify a system that has undefeated Alabama ranked lower than a team they beat on a neutral field this season?!

by TV_Pete (not verified) :: Thu, 10/08/2009 - 10:35am

I agree, on a subjective basis, that Alabama is a more complete team.

However, this is a computer system based upon drive outcomes. Is it very positive, positive, neutral, negative, or very negative. Only drives that are "relevant" are counted. I think competition against 1-AA teams (FCS?) are not counted. For the next week or two, preseason weighting is still included, but it will be dropped after that.

I don't think this system is the perfect answer. However, I would consider in my weights (as one factor in my own informal BCS ranking) and it is not subjective. (I would include my personal weight and Polls for subjective addition, which can measure factors that may not be picked up, such as Bradford's injury and return, etc.)

by xLittleP :: Thu, 10/08/2009 - 8:47am

Georgia Tech gets knocked out of the top 25? Well, our defense is terrible... (Hey maybe the game should be trying to come up with a sentence or a template that explains why the computer is right?)

Looking at FSU's efficiencies though, they are worse than GT in both aspects. Brian, when your predictions come out, how do you account for home field advantage?

Also, since you bought up SOS, how do you calculate an "top-5 team"? Is it the average of the top five teams in the current FEI, or is it the fifth place team? I'm just curious.

I love the FEI, btw, it usually is very kind to the Jackets, haha.

by tabsports :: Thu, 10/08/2009 - 12:49pm

I'm confused. TCU is ahead of Alabama, yet Alabama has a better game efficiency, strength of schedule and FEI projection by a healthy margin.

We may have someone whining about VaTech over Alabama, but that's understandable (tougher schedule so far and higher projection). Meanwhile, I'm completely at a loss of understanding TCU over Alabama (or TCU in the top ten, for that matter). No beef that TCU is there, just don't understand why, given the basics in this chart and the preseason projections.

by Corey (not verified) :: Thu, 10/08/2009 - 7:47pm

Exactly. This stuff is way out-of-whack based upon the data presented. Is the preseason projections still that much of a facotr in these calculations?

by Brian Fremeau :: Fri, 10/09/2009 - 4:04pm

Sorry if I haven't answered questions directly on some teams, but it looks like I need to clear up some more confusion on the process.

The individual GE, OE, DE and FPA numbers in the charts represent games played to date. The SOS numbers in the chart represent each team's overall 2009 schedule, including games not yet played. The FEI numbers in the chart represent a combination of games played to date (GE adjusted for strength of opposition faced) and projected data.

Nebraska had a projection in the 60s to start the year (Pitt's was No. 30). TCU is an odd case -- I'll have to look closely at what's happening there, but I think it has something to do with wins over Clemson and Virginia and the preseason projections of those two teams. In other words, it's probably temporary.