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?

05 Nov 2008

FEI Week 10 Ratings

by Brian Fremeau

The Fremeau Efficiency Index 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. Strength of Schedule (SOS) is calculated from a privileged perspective (explained here) and represents the likelihood that an Elite team (top 5) would post an undefeated record against the given team's opponents to date.

The following ratings are calculated based on data from all FBS games played through Sunday, November 2. Only games between FBS teams are considered.

Rank Team Record FEI Last Week vs. Top 10 vs. Top 40 GE GE Rank SOS SOS Rank
1 Florida 7-1 0.303 5 0-0 2-1 0.414 2 0.415 51
2 Penn State 8-0 0.293 2 0-0 2-0 0.389 3 0.548 72
3 Texas 8-1 0.288 1 1-1 3-1 0.349 5 0.248 15
4 North Carolina 5-2 0.263 3 0-0 3-2 0.161 17 0.345 33
5 USC 7-1 0.255 4 0-0 2-0 0.427 1 0.562 75
6 Alabama 9-0 0.238 6 0-0 2-0 0.318 8 0.504 65
7 Texas Tech 7-0 0.237 12 1-0 1-0 0.353 4 0.460 57
8 Georgia Tech 5-2 0.223 9 1-0 3-2 0.126 21 0.344 32
9 Oklahoma 7-1 0.218 10 0-1 2-1 0.323 7 0.380 42
10 Florida State 4-2 0.190 13 0-1 2-2 0.068 34 0.320 27
11 Virginia Tech 4-3 0.182 8 2-1 2-3 0.023 52 0.218 9
12 Oklahoma State 7-1 0.173 11 0-1 1-1 0.279 11 0.362 36
Rank Team Record FEI Last Week vs. Top 10 vs. Top 40
13 Ohio State 6-2 0.172 15 0-2 2-2 0.088 26 0.195 7
14 Mississippi 4-4 0.171 17 1-1 1-4 0.036 49 0.154 4
15 Missouri 6-2 0.166 14 0-1 1-2 0.205 15 0.309 25
16 Georgia 6-2 0.156 7 0-2 2-2 0.041 47 0.176 6
17 Miami 5-3 0.154 21 0-3 3-3 0.033 50 0.125 3
18 Wake Forest 5-3 0.145 24 1-0 3-1 0.011 57 0.324 29
19 Vanderbilt 5-3 0.144 18 0-0 2-2 0.071 33 0.412 50
20 Ball State 7-0 0.143 20 0-0 0-0 0.327 6 0.852 119
21 Pittsburgh 6-2 0.143 22 0-0 1-0 0.071 32 0.561 74
22 Iowa 4-4 0.133 23 0-0 0-4 0.137 20 0.450 55
23 Northwestern 6-2 0.129 33 0-0 3-1 0.061 36 0.502 64
24 Michigan State 8-2 0.128 27 0-0 2-2 0.089 25 0.393 44
25 Connecticut 5-3 0.123 16 0-1 2-2 0.045 44 0.370 38

Adjusted Offensive Efficiency (AOE) and Adjusted Defensive Efficiency (ADE) are the opponent-adjusted values of Offensive Efficiency (OE) and Defensive Efficiency (DE), explained here. Like FEI, the multiple-order adjustments are weighted according to both the strength of the opponent and the relative significance of the result; efficiency against a team's best competition faced is given more relevance weight. AOE and ADE represent a team's value over/under average. Positive AOE and negative ADE are the most valuable.

Rank Team Record AOE AOE Rank ADE ADE Rank OE OE Rank DE DE Rank
1 Florida 7-1 0.411 7 -0.460 1 0.507 13 -0.587 3
2 Penn State 8-0 0.511 3 -0.405 5 0.539 12 -0.531 6
3 Texas 8-1 0.586 1 -0.308 16 0.963 1 -0.125 42
4 North Carolina 5-2 0.185 32 -0.441 3 -0.048 61 -0.359 14
5 USC 7-1 0.424 6 -0.376 9 0.566 9 -0.640 1
6 Alabama 9-0 0.373 8 -0.389 7 0.229 29 -0.630 2
7 Texas Tech 7-0 0.370 9 -0.361 11 0.940 2 -0.046 56
8 Georgia Tech 5-2 0.303 15 -0.217 27 0.023 48 -0.329 20
9 Oklahoma 7-1 0.540 2 -0.229 24 0.892 4 -0.097 48
10 Florida State 4-2 0.147 37 -0.187 35 -0.039 60 -0.173 37
11 Virginia Tech 4-3 -0.011 66 -0.302 18 -0.218 82 -0.111 46
12 Oklahoma State 7-1 0.306 14 -0.212 28 0.676 8 -0.058 54
Rank Team Record AOE AOE Rank ADE ADE Rank
OE OE Rank DE DE Rank
13 Ohio State 6-2 0.057 53 -0.405 4 -0.177 73 -0.414 9
14 Mississippi 4-4 0.293 17 -0.223 26 -0.010 51 -0.121 44
15 Missouri 6-2 0.438 5 -0.062 60 0.755 6 0.136 83
16 Georgia 6-2 0.330 11 -0.162 41 0.117 40 -0.095 49
17 Miami 5-3 0.070 50 -0.004 66 -0.198 77 -0.071 52
18 Wake Forest 5-3 -0.248 96 -0.441 2 -0.367 107 -0.224 33
19 Vanderbilt 5-3 -0.020 69 -0.400 6 -0.211 81 -0.317 22
20 Ball State 7-0 0.285 19 -0.250 21 0.769 5 -0.254 27
21 Pittsburgh 6-2 0.307 13 -0.250 21 0.769 5 -0.254 27
22 Iowa 4-4 0.009 61 -0.385 8 -0.037 59 -0.465 7
23 Northwestern 6-2 0.273 21 -0.169 39 -0.029 56 -0.237 30
24 Michigan State 8-2 0.158 35 -0.144 44 -0.069 64 -0.157 39
25 Connecticut 5-3 0.084 48 -0.335 14 -0.026 54 -0.304 23

The Week 10 FEI Ratings for all 120 FBS teams can be found here. Expanded FEI Ratings data can be found here.

Florida takes over the top spot in the Week 10 FEI Ratings after stomping Georgia in Jacksonville. The Gators are the seventh team to hold the FEI No. 1 spot this season, a position that has changed hands five times in the last six weeks. Six weeks ago, we were eagerly anticipating the first game of five regular season match-ups between the four SEC "powerhouses" -- Florida, Georgia, LSU, and Alabama -- all ranked in the FEI top 8 through week four of the season. Four of those five games are now in the books, none of them were particularly close, and the Gators currently stand atop their mini-tournament standings at 2-0, blowing out both the Bulldogs and Tigers. Alabama will have their opportunity to match the Gators' feat this Saturday in Baton Rouge, and will likely face Florida for the SEC title in December.

No. 3 Texas lost the game of the year in Lubbock Saturday night, and has mercifully closed its own mini-tournament, going 3-1 in four straight weeks against the FEI No. 7 (Texas Tech), No. 9 (Oklahoma), No. 12 (Oklahoma State), and No. 15 (Missouri) teams. Among the top five teams in the Big 12, eight total regular season games will ultimately be played this year. Of the five in the books, only one was decided by more than a single possession, Texas' 56-31 obliteration of Missouri on October 18.

The Red Raiders' thrilling win over the Longhorns launched them into the BCS title game conversation, but it was Texas Tech's first game against an FEI top 40 opponent this season. The Big 12 South driver's seat is theirs, but after a brutal month, Texas is probably happy to hand over the keys. Riding shotgun might be the best BCS position in early November.

A few weeks ago, I calculated the likelihood that each of the ten then-undefeated teams would run the table through the remainder of the year. Let's revisit that table, including the six remaining undefeated teams plus the seven one-loss teams waiting at the BCS doorstep. The likelihood is a function of the Projected Win Expectation of each team in each match-up, based on the current FEI ratings for each team. Teams listed with an asterisk may possibly also play a conference championship game at season's end. Those possible games are not factored in the table.

Likelihood of Winning All Remaining Regular Season Games
FEI Rank Team W-L Likelihood Pct. Games Remaining BCS Rank
3 Texas* 8-1 95.0% Baylor, Kansas, Texas A&M 4
26 Boise State 7-0 78.4% Utah State, Idaho, Nevada, Fresno State 10
6 Alabama* 9-0 73.5% LSU, Mississippi State, Auburn 1
5 USC 7-1 72.7% California, Stanford, Notre Dame, UCLA 7
2 Penn State 8-0 69.5% Iowa, Indiana, Michigan State 3
1 Florida* 7-1 55.5% Vanderbilt, South Carolina, Citadel, Florida State 5
34 TCU 8-1 37.9% Utah, Air Force 12
7 Texas Tech* 7-0 37.0% Oklahoma State, Oklahoma, Baylor 2
20 Ball State* 7-0 36.5% Northern Illinois, Miami (OH), Central Michigan, Western Michigan 17
37 Utah 8-0 34.1% TCU, San Diego State, BYU 8
9 Oklahoma* 7-1 26.8% Texas A&M, Texas Tech, Oklahoma State 6
62 BYU 7-1 12.3% San Diego State, Air Force, Utah 15
12 Oklahoma State* 7-1 11.7% Texas Tech, Colorado, Oklahoma 9

The five teams in the best position three weeks ago to remain undefeated have done exactly that. Texas Tech has improved its position, but is far from a sure thing the rest of the way, and Texas can probably count on some of the dominoes they need to fall. Penn State and USC are good bets to hang around while others struggle as well.

The six remaining undefeated teams don't play one another, so it is mathematically possible for all of them to run the table, however unlikely. The 13 teams listed above do play five more regular season games against one another, plus a probable sixth in the SEC championship. In short, even without dropping a game to someone outside the BCS top 15 (the group has already suffered two such defeats this year), the 13 teams listed above will have at least a combined 13 losses before bowl season.

As my colleague Russell Levine discussed in his column this week, projecting, dreading, and/or eagerly anticipating BCS chaos is the pastime of choice this time of year. Like Russell, I'm always surprised that the speculation usually focuses more on poll position than remaining schedule. An undefeated Penn State team could indeed be shut out of a title game opportunity if Texas Tech and Alabama also win out, but FEI projects only a 18.9 percent likelihood of that situation occurring. If undefeated Red Raiders and Crimson Tide teams meet Missouri and Florida in their respective conference championship games, the likelihood that all three current BCS conference undefeated teams remain unblemished prior to bowl season plummets to about 5 percent.

If only two BCS-conference teams remain undefeated and play for the championship, will chaos have been averted? In 2005, USC and Texas fit the bill and played an epic thriller for the national title, the only game played between the postseason FEI No. 1 and No. 2 teams since 2003. I am confident that there will be arguments this year, and we may indeed be careening toward an unprecedented debate over the merits of teams, conferences, and even conference divisions. I am less confident, but still hopeful, that the teams that do ultimately appear in the BCS championship game can at least match if not exceed the drama in Lubbock on Saturday night.

Posted by: Brian Fremeau on 05 Nov 2008

26 comments, Last at 06 Nov 2008, 4:57pm by Pat (filler)


by Pete (not verified) :: Wed, 11/05/2008 - 12:00pm

Go, Gators! :-)

After the loss to Mississippi I felt like they were playing like a Top 10-Top 15 team. They won games before then by a nice margin, but it took most of the game to put away Miami and UT. Since that game I think Tebow and the coaches have motivated the Gators to play like the best team in the country.

After the win over Georgia can they continue to play at that level or will they ease up?

If I were to suggest improvements for the BCS:

1) Do not force computer rankings (Sagarin, specifically) to ignore margin of victory. I like Sagarin rankings (or predictor, I guess) more than the ELO-Chess that is forced upon us.

2) Include FEI as one of the computer rankings (heh). I doubt this is going to happen due to the potential for Fremeau bias (what is definition of "garbage time" or "successful drive", lack of automation, etc.)

3) Run a classic Bowl + 1. Most of the "Plus One" suggest organizing a 4-team playoff at two of the Bowls. This has a lot of resistance from the PAC-10 and Big 10. Instead, let the class Bowl matchups do their thing (SEC to Sugar Bowl maybe facing an undefeated non-BCS team, normal Rose Bowl probably between USC and Penn State, etc.). Then, after there is some guaranteed quality inter-conference play, run the numbers through the BCS and determine who will play in the championship. The current championship is about a week later, so it would not need to extend the timing (much, if any). Only two teams would play an additional game.

by Pat (filler) (not verified) :: Wed, 11/05/2008 - 4:09pm

1) Do not force computer rankings (Sagarin, specifically) to ignore margin of victory. I like Sagarin rankings (or predictor, I guess) more than the ELO-Chess that is forced upon us.

Nooooo!! Dear God, no!

There are very, very good reasons why margin of victory is excluded from statistical rankings. The simplest reason is that it reduces the value of a win to zero in the limit of a close game. The more complicated reason is that there is no unbiasable measure of victory in football.

The reason that human polls are 2/3rds of the BCS is to compensate for the limitations of the statistical rankings. Using margin of victory is a very bad idea.

Elo rankings are well known and very popular worldwide. They're used in several other sports and competitions. I personally don't think they're very well suited for college football (they converge too slowly, and there aren't enough losses in the Top 25) but the basic idea is sound.

by Dave R (not verified) :: Wed, 11/05/2008 - 5:00pm

There might be good reasons for leaving out margin of victory from the formula, but computer rankings that use margin of victory seem to be more accurate, and it was eliminated from the BCS largely for emotional reasons (because they thought it would encourage running up the score, which it wouldn't, because the margin of victory effect was capped in all the BCS computers that used it).

by Pat (filler) (not verified) :: Wed, 11/05/2008 - 6:11pm

I'll try to keep this simple and short. Predictive accuracy doesn't matter in a ranking system used to determine a champion. More than anything, you need a descriptive (i.e. retrodictive) and bias-free ranking. You want a descriptive ranking because you are judging teams on what they have done, not what they will do. You want a bias-free ranking, because with such small samples, you can't afford to make a mistake on even one game, since one game might carry all the weight.

It's pretty simple: you don't use margin-of-victory in football because football isn't baseball. Baseball has a (nearly unbiasable) descriptor of the degree to which one team wins: margin of victory. Scoring runs doesn't ever decrease your chances of winning, with the slight exception of pinch-hitting.

Football is a timed game, however, and so scoring points can, and does decrease your chances of winning. A team with less than 1 minute to play, up 1 point, from the 20 yard line, should not kick a field goal, or try to score a TD. They should kneel three times. Kicking the field goal would heavily decrease their chances of winning.

Now ask yourself this: when do these problems show up? In close games. What games do people consider the most important in college football? Games between good opponents. Which tend to be close.

The entire point of the human polls is to deal with the fact that the statistical rankings can't handle degree of victory. It's fundamentally impossible, and nothing says it better than this: a game won 13-6 in triple-overtime has a bigger margin of victory than a game that was 13-0 for the entire game before a hail-mary pass on the last play of the game made it 13-7.

by Pete (not verified) :: Wed, 11/05/2008 - 11:34pm

Georgia and Florida are good opponents (as might be LSU). However, if a team CAN beat their opponent by 17+ (3 scores) points then they should try to do that. At the end of a close game, yes, the best choice is usually to run out the clock. However, which is a more decisive victory? Penn State beating Ohio 13-7 (when a single call or non-fumble might have resulted in a loss by Penn State) or Florida beating Georgia 49-10? The Keg (not cocktail) Party was much closer than that score, but Florida was playing like a much better team.

In one instance of the BCS the record counted for a portion (20%, I think) and the 5 or 6 computer rankings (only 2 or 3 included margin of victory) and they all used a diminishing returns (anything over 20 points is pretty negligible).

You may notice that I prefer the Sagarin Rankings, which is a merge of his predictor (his predictor for which team is the best and which team will win, seems pretty useful to me) and the ELO-Chess (which is my least favorite of the three).

The Coaches Polls are not reported (who picked which team where), are biased (rank your team and opponents high and it benefits you), and may not even be filled out by the Coaches (and if filled out by Coaches they may not have seen as many games as those of us here). However, the human polls can take into account some things that an unbiased computer ranking cannot (injuries, vendettas, good match up, suspended or returning players).

by Pat (filler) (not verified) :: Thu, 11/06/2008 - 12:05am

If you have a problem with the human polls, complain about them. Why are you suggesting a flawed system be implemented to fix a differently-flawed system? Wouldn't the correct solution be to implement a non flawed system?

However, if a team CAN beat their opponent by 17+ (3 scores) points then they should try to do that.

No, it shouldn't. That's fundamentally wrong football. Once you've got a lead, near the end of the game, the ideal situation is to run out the clock, not try to score. Look at the Penn State-Illinois game for an example - Penn State had the ball in easy field goal range, and they ran the ball to kill the clock. So they won by 14, instead of by 17. But they made the right decision.

and they all used a diminishing returns

I really think you're missing the point. It's not the fact that there are diminishing returns on margin-of-victory. It's the fact that you cannot weight it that strongly. You can't tell the difference between a strong 14-point lead which turned into a 7-point win late, and a close 7-point win that happened late.

Which means that realistically, margin of victory is only really helpful if you can use huge margins. Because if you cap it at 15, and you can't tell the difference between a 7 point win and a 14-point win, what the heck is it good for? This is why Billingsley (insane as he is) stripped it from his rankings well before the BCS asked them to.

On average margin of victory tends to work. That's because the outlier situations (where the score changes late) are rare. But you are not ranking teams on average games. You're ranking teams based on their actual games.

by navin :: Wed, 11/05/2008 - 10:45pm

Question for you since you're very high on pure win/loss for college football.

What if Texas Tech only loses to Oklahoma, and we finish with Texas, Texas Tech, and Oklahoma all with one loss. Could you make a convincing argument for ranking an undefeated Penn State above all of the three?

Normally the loss connects a team to the the rest of the field, but not so in this case. I'm interested in hearing the math behind the argument.

by Pat (filler) (not verified) :: Thu, 11/06/2008 - 12:36am

Could you make a convincing argument for ranking an undefeated Penn State above all of the three?

The circle of death essentially equalizes TT/UT/OU, so start by taking away all those games, and treat the teams as unbeatens with the rest of their schedule. The math of a maximum-likelihood ranking would then "tug" the teams together. Texas and Oklahoma would be ranked close, and Texas Tech much lower.

Texas Tech's schedule (without OU/UT on it) is so weak that yeah, in all likelihood, Penn State would be ranked above them.

The mechanisms for ranking unbeaten teams are really sketchy though, even with margin-of-victory. The idea that you suddenly "fix" a system that's just totally broken (way too few games between teams of interest) by adding in a few numbers that are nigh-on impossible to interpret is just goofy.

Basically, the problem's this: when you've got an unbeaten, you have to fix it by asking this: "how likely is it that a team is this good?" Even if you've got margin-of-victory, it's the same problem - especially if you cap margin of victory. If you cap margin of victory, and the team beats every team by that cap or more, how do you rank them? An infinitely good team would've done just as well. The problem with asking the "how likely is it.." question is that you're assuming that it's just as likely for a team like Ball State to be a Top 3 team as it is for Penn State, Texas, Alabama, etc. Which means you end up boosting midmajors and weakening major teams.

by Dave R (not verified) :: Wed, 11/05/2008 - 4:52pm

Classic Bowl +1 is about the only psuedo-playoff format that I really hate; it's rarely an improvement over the current system. If we decide Big East/ACC is a 'classic' matchup for the Orange, and use the current BCS rules for two at-large spots rather than 4 (i.e. first a non-BCS in the top 12/above the top BCS champ, then ND if in the top 8, then #3 or #4; after that, I used the highest-ranked teams that wouldn't match up teams from the same conference), then backtracking a bit (and assuming that the highest-ranked team in the final BCS rankings won their conference, unless I remember otherwise, because I'm too lazy to look that up)...

in 2007, you'd likely have
Rose: #7 USC beats #1 Ohio State
Fiesta: #4 Oklahoma beats #11 Hawaii
Sugar: #2 LSU beats #8 Kansas/#6 Missouri (maybe; this is the only game that would match two teams that won their actual bowl games)
Orange: #9 West Virginia beats #3 Virginia Tech

At that point, who plays in the title game? USC just beat #1, but Oklahoma, LSU, and WVU also just beat highly-ranked teams, and all have the same record.

Rose: #5 USC beats #1 Ohio State
Fiesta: #8 Boise beats #10 Oklahoma (actually happened, or we wouldn't project this)
Sugar: #2 Florida beats #3 Michigan (maybe; non-Ohio State Big Ten teams have done pretty well against the ACC)
Orange: #6 Louisville beats #13 Wake Forest (actually happened)

Here, Florida and Louisville have the best records among BCS conference teams, Boise is undefeated and just beat Oklahoma, and USC just knocked off #1.

Rose: #1 USC beats #3 Penn State
Fiesta: #2 Texas beats #4 Ohio State (regular season rematch)
Sugar: #7 Georgia beats #6 ND
Orange: #11 West Virginia beats #10 Virginia Tech

So in the first case where you've got a clear title game, you get exactly the same USC-Texas game you would have originally had.

Rose: #1 USC beats #11 Iowa
Fiesta: #6 Utah beats #2 Oklahoma
Sugar: #3 Auburn beats #4 Texas
Orange: #16 Florida State beats #21 Pitt

In this season, you end up with two undefeated major conference teams, and an undefeated mid-major that just knocked off the #2 team in the country. You probably get a USC-Auburn title game, but the Utes would be... upset.

Rose: #3 USC beats #4 Michigan
Fiesta: #5 Michigan beats #10 Kansas State
Sugar: #2 LSU beats #1 Oklahoma (actually happened)
Orange: #9 Miami beats #7 Florida State (I think this would be a rematch, in Miami's home stadium)

So here's the first time where this format actually helps, as we can have our USC/LSU game (though the expectation is that USC would have won it handily).

Rose: #2 Ohio State beats #6 Washington State
Fiesta: #4 USC beats #7 Oklahoma
Sugar: #3 Georgia beats #5 Iowa
Orange: #1 Miami beats #14 Florida State

So you get Ohio State-Miami again, with a nagging impression that if either played USC, they'd get killed... which is what really happened.

Rose: #4 Oregon beats #8 Illinois
Fiesta: #3 Colorado beats #6 Tennessee
Sugar: #5 Florida beats #2 Nebraska (probably, anyway)
Orange: #1 Miami beats #10 Maryland

And, well, you'd think that would get a Miami/Oregon title game (because using the current BCS formula, Oregon would be #2 going in), so maybe this is the second time it helps.

Not a great record.

by Pete (not verified) :: Wed, 11/05/2008 - 11:40pm

What is the point of the BCS rankings? It is not to find a champion. It is to find a number of teams (2 in this case) to compete for a championship. Theoretically, it is used to find the best teams in the country.

Adding a quality game before that determination is made seems like it would be worthwhile to me. It may not be clean and may result in more teams competing for championships with at least 1 loss, but I think it is a step in the right direction.

The past two years (and this one) I have not believed that Ohio State was one of the 2 best teams in the country. However, since they were undefeated on December 12th they got to go to the Championship. If they played USC the past few years I expect they would have been revealed as Top 5-10 teams rather than one of the 2 best.

The BCS rankings would then determine which 2 teams would compete for the championship (probably USC beating LSU in 2007, although as an SEC homer I would root for LSU and Georgia).

by CuseFanInSoCal :: Thu, 11/06/2008 - 3:52am

The thing is that an additional tough game in anything other than a seeded playoff is as likely to create ambiguity as reduce it. Which I think I showed (Dave R being me before I registered).

My new CuseFanInSoCal blog

by sethburn :: Wed, 11/05/2008 - 12:51pm

It is important to remember that Alabama's chances of going undefeated through the regular season are much higher than their likelihood of beating Florida in the SEC championship game. If Florida has only one loss and Alabama is undefeated that will be a monster game. Right now the most likely national championship game looks to be Florida vs. Penn St. due to the expectation that Florida beats Alabama and Oklahoma beats Texas Tech.

by navin :: Wed, 11/05/2008 - 10:42pm

Texas deserves to play in the title game if they win the Big 12 and Alabama loses.

Losing on the road to Texas Tech>>>>>>>Losing at home to Ole Miss.

by War Eagle (not verified) :: Wed, 11/05/2008 - 3:22pm

I'm surprised, to be honest, that the consensus seems to be so clear that Florida will beat Alabama. What Florida is good at doing (inside running), Alabama is quite good at stopping. Further, the non-Harvin receivers are quite average. I also think that Bama can probably run the ball against anyone with that line and stable of backs. We'll see I guess.

by Pete (not verified) :: Wed, 11/05/2008 - 3:38pm

I thought what Florida was better at (at least with the 1-2-3 Running backs) is running option-sweeps and off-tackle rather than up the middle. Most of the year they have been throwing dink-and-dunk passes (mostly to Harvin, although they try to throw to Murphy who likes dropping or giving up on passes). Tebow has been able to complete a few passes where he throws more than 12 yards in the past couple games, so that may be promising.

However, I agree that Alabama's DL is the best that Florida will have faced. When Tebow is pressured he has been less efficient and fumbled or (almost) thrown interceptions.

by sam :: Wed, 11/05/2008 - 4:03pm

Regardless of what Florida "does best," their offensecan beat you a lot of different ways. Teblow spreads the ball around and the plan seems different every game they play. Their defense has also been improving every week.

sam! or the original sam from the old FO

by Pat (filler) (not verified) :: Wed, 11/05/2008 - 4:18pm

Incidentally, regarding the whole ACC thing:

The ACC is the worst-connected conference in college football this year, by a fairly large margin: they have 14 games against Division IAA opponents. No other conference has more than 10. And, in addition, several of their non-conference games still have yet to be played - GT/UGA, UF/FSU, for instance.

Unconnected conferences tend to "spread" to the top and bottom of any opponent-adjusted ranking: if you think about two disconnected groups, the highest ranked teams would likely be the same as one another, and the same for the lowest ranked teams (more or less - obviously using performance rather than just wins/losses ties things a little, but not that much).

It could just be a coincidence, but it's a bit suspicious that a conference that's that poorly connected to the rest of college football is considered that strong.

by War Eagle (not verified) :: Wed, 11/05/2008 - 4:33pm

I've watched a fair amount of Florida games and when they've been efficient, they're running their veer, and basically overpowering teams at the point of attack. Then they start mixing in the outside runs and more complex passing game. They've been fairly inefficient when teams have taken away their inside runs, in my opinion.

They have a young offensive line that struggled early in the year. They went simple, with lots of running plays (easier than pass blocking), especially between the tackles, and it has worked. Nobody has really taken that away in the last few games. I think Alabama certainly could. I've not been very impressed with Florida's passing offense this year, and I'd be pretty tempted to make them beat me with that.

by Kibbles :: Wed, 11/05/2008 - 8:05pm

You're focusing on how Florida's offense is going to score on Alabama's defense. I've got a better question for you- how the heck is Alabama's offense going to score on FLORIDA'S defense? Alabama has a worse offense than Florida with far fewer "strengths", and Florida has a significantly better defense than Alabama.

Back to your original question, though, of how Florida would score on Alabama... pretty easily, I'd imagine, what with all the short fields they'd be facing after those long Brandon James returns, the glut of turnovers, and the odd blocked kick. Pretty much the same as they have all year. Their average drive against Georgia started at midfield. I think any defense would be hard pressed to consistently keep Tebow, Harvin, Demps, Rainey, and company out of the end zone on a short field, regardless of their particular strengths.

by navin :: Wed, 11/05/2008 - 10:41pm

How did Ole Miss score against Florida?

This week should be helpful. If Alabama can score on LSU, then they'll be fine. Otherwise they had better fix their offense.

by Pete (not verified) :: Wed, 11/05/2008 - 11:24pm

Alabama, like Georgia, got a few nice runs up the middle (against a young set of DT). However, their most consistent successes were taking advantage of the DE who lined up over the DT and did not maintain containment around the edge. If the WR blocks the CB (or the CB takes himself out of the picture) then a big gain is possible for a quick RB/WR.

There was a big pass, but that was really a single play.

Mostly, I thought 14 points resulted from an aggressive defense fighting to get 3 out of the 5 fumbles.

Tebow seems to be passing better now, but it would be interesting to see a team try to make him beat them (especially with tight coverage against Harvin, if possible).

by lionsbob :: Wed, 11/05/2008 - 10:48pm

Alabama is going to line it up and shove down Florida's throat if they play in the SEC championship.

Alabama's biggest problem is that they have a true freshman #1 WR and nobody has really stepped up to be the #2 guy. We have a bunch of complimentary WRs, but not a true #2 (though a healthy Mike McCoy might be able to fill in).

by Brass Rat (not verified) :: Wed, 11/05/2008 - 11:31pm

On the topic of whether the computers should include MOV, Pat makes the strongest argument for it when he acknowledges that Sagarin's Elo converges too slowly. With only 12 regular season games, and typically only 1 or 2 non-confernce games per team that are at all helpful in providing inter-conference pecking orders, the computers simply don't have enough data with win/loss only. A great example of this was Virginia Tech being ranked #1 by several BCS computers at the end of 2007. The computers could not factor the blowout loss that Va Tech had against LSU. Sagarin's Elo Chess had Va Tech #1, but his MOV (or in this case, margin of defeat) rankings had them at #5, a more appropriate spot. Georgia should be penalized more for getting blown out by Florida and Bama, as opposed to if they had been close games. And while I agree that what you accomplish is retrodictive, accurately assessing the strength of your opponents should place greater emphasis on their "predictive" ranking. What I mean by this is that if you play a good team that has underperformed by blowing a game, the assessment of your schedule should reflect the strength of the team, not their retrodictive accomplishment. And MOV (and MOD) allows the computers a much better ability of assessing that strength.

by Pat (filler) (not verified) :: Wed, 11/05/2008 - 11:44pm

Elo converges slowly because it's designed to converge slowly. It's a stepwise convergence to a maximum likelihood point. If you just either do the math or have some sort of insane fitter find the maximum, you'll do much better.

accurately assessing the strength of your opponents should place greater emphasis on their "predictive" ranking.

Hence the reason for a human poll. You cannot do it correctly with statistical rankings. The fact that you're cherrypicking situations where you "liked" the result doesn't change the fact that from a fundamental point of view, it's just fatally flawed.

What do you do when the blowout loss by Virginia Tech happens because of an injury that's gone by the next week? Since you have to cap the scoring margin at, say, 15 or so, how do you tell the difference between a 15-point victory that happened at the last moment and a 40-point victory that was held throughout the second half? Humans can filter that information. A statistical poll can't.

by Becephalus :: Thu, 11/06/2008 - 8:03am

So why not commission a committee of experts tasked with watching all the important games and making a decision? Instead of relying on polls of two groups on incredibly partisan and badly informed individuals. I personally think a large part of the problem with the reputation of human pools is that they are polling the wrong people.

The Wire should win the Nobel prize for literature.

by Pat (filler) (not verified) :: Thu, 11/06/2008 - 4:57pm

That's fine by me. I have no special love of the Coaches or Harris polls. I just know that attempting to fix a system that's sporadically biased with one that's consistently, and provably biased is a horrible idea.