Sunday, September 12, 2010

Week 3: Top 25 -- TFG

Another week in the books, some more churn in the top 25. Expect to see more of this up-and-down until the rankings really settle into place for this season.

Rank +/- Team WinPct SoS Off Pts Def Pts Pace
001-- Alabama 0.9529 0.6167 28.5 7.8 79.1
002-- Florida 0.9212 0.5610 30.6 10.9 78.2
003-- Ohio St. 0.9105 0.5661 24.8 9.2 82.1
004-- TCU 0.9076 0.5255 24.8 9.5 83.0
005+1 Oregon 0.8962 0.5761 26.6 10.8 90.1
006-1 Texas 0.8867 0.4797 28.0 11.8 82.8
007-- Boise St. 0.8713 0.5073 27.3 12.2 83.7
008+2 Oklahoma 0.8676 0.6303 22.3 10.1 89.8
009-- Virginia Tech 0.8637 0.6456 26.8 12.3 79.5
010+1 Iowa 0.8499 0.5443 21.7 10.4 81.0
011+2 LSU 0.8362 0.5894 22.6 11.3 81.6
012-- Nebraska 0.8140 0.4252 20.6 11.0 80.9
013-5 Penn State 0.8125 0.6722 20.5 11.0 79.4
014NA Stanford 0.7925 0.5834 27.6 15.6 81.2
015-1 USC 0.7855 0.5661 21.1 12.2 85.1
016+1 Utah 0.7759 0.4664 21.8 13.0 84.2
017-2 Texas Tech 0.7632 0.4955 24.8 15.3 89.2
018+7 Arkansas 0.7607 0.5809 25.1 15.4 83.2
019+4 Arizona 0.7507 0.5525 21.3 13.5 82.5
020NA Air Force 0.7385 0.4764 19.9 12.9 83.3
021-2 Clemson 0.7352 0.5100 21.0 13.5 81.5
022NA Miami-FL 0.7283 0.6790 22.7 15.0 83.0
023+1 Oregon St. 0.7264 0.6759 24.0 15.9 84.4
024NA South Carolina 0.7185 0.6374 17.9 12.1 80.7
025-9 Georgia 0.7156 0.5978 24.2 16.4 78.6

New entries: Stanford, Air Force, Miami-FL, South Carolina.

Dropped out: BYU, Mississippi, Georgia Tech, Pittsburgh.

Even in defeat, the Hurricanes managed to leap four spots into the top 25 by hanging 24 points on a tough Buckeye defense. Granted, 14 of those points came from special teams when OSU's defense was not on the field, per se, but TFG doesn't distinguish between offense, defense, and special teams.  The other three new entries all leap into our top 25 thanks to some solid wins, including South Carolina defeating Georgia in our inaugural GUGS Game of the Week.

At the top we see Alabama opening a decisive gap between themselves and second-place Florida. The Tide gets major points for holding the Nittany Lion offense to 3 points, making themselves not just the only team below the 9.0 PPH mark on defense, but below 8.0 PPH. So far this year, Alabama has stepped up their play on offense, breaking into the 28 PPH level, second only to Florida. Florida has seemed shaky at times this year, so SEC play will really shed some light on whether these unreal efficiencies are accurate, or an artifact of playing inferior competition (plus a bad day by Penn State).

Right now all the hype appears to be about Alabama and Ohio State, but the gap between (1) and (3) is larger than the gap between the third-ranked Buckeyes and the 7th-ranked Boise State Broncos. Even without taking Virginia Tech's loss into account, Boise State has yet to really differentiate themselves from the rest of the middle of the top 10.

Further down the rankings we see USC feeling the effects of their diminished payroll, currently ranked as the third-best team in the Pac 10, with Arizona closing fast. The Pac10 itself may be the sleeper story of the year so far, with half of their teams in the top 25. TFG has them listed as the second-best conference in college football, behind only the SEC. The Massey Ratings, a comparison of dozens of computerized college football rankings in which both TFG and RBA are listed, has the Pac10 as the third-best conference behind the SEC and the Big 12. We'll see how well that holds up as the year goes on, but so far this year the Pac10 has broken out of the traditional "USC and the 9 dwarves" storyline.

Saturday, September 11, 2010

Week 2: Saturday Predictions

Full predictions after the jump. Justin's picks are in blue, Eddie's are in yellow.

Week 2: Saturday Matchups

Here at TFG we're committed to bringing our readers a slightly different perspective on college football.  In light of that, we're working on our own metric to predict which games will be the most "interesting" in any given week. Ladies and gentlemen, we give you the Games You Gotta See (GUGS) system. This was inspired by Ken Pomeroy's FanMatch system, the exact inner workings of which are a bit of a mystery. We previewed this during our bowl coverage last year when we attempted to rank all 34 bowl games from least to most interesting.

The major factors that make a game "interesting", in our opinion, are the combined quality of the teams and the predicted closeness of the game. Pomeroy's FanMatch also gives weight to how high-scoring the game will be; we haven't included that, but probably will at some point.  For now we give some weight to team quality, and more weight to predicted closeness. Our view of what's interesting might not always match up with what the rest of the sports world sees as interesting. For example, most sports pundits are talking up Penn State-Alabama and Ohio State-Miami (FL) this weekend; we can't deny that these games have national title implications and include four of the most storied teams in college football history. However, our system is less impressed with these games for one simple reason: they aren't expected to be close.

Let's look at the TFG Games of the Week, and then we'll come back to a few big games this weekend that failed to twiddle our bits.

Georgia Bulldogs at South Carolina Gamecocks.
GUGS Score: 85.7

Eddie:  RBA ranks (12) Georgia higher than (15) South Carolina, so you'd expect them to be favored in this matchup.  However, the game is in Columbia, so add (approximately) three points to South Carolina.  Georgia looked very impressive last week against Louisiana-Lafayette, but that only serves to verify their tendency to whip up on weak competition (expected 36.1 PPH against LA-Lafayette).  Historically, Georgia struggles against higher quality opponents.  Their offensive efficiency is expected to drop by 56% to 15.9 PPH.  South Carolina exploded on Southern Miss last week but trend at 11.2 PPH versus teams of Georgia's strength.  The real issue in this game is the defensive battle -- Georgia should yield 19.2 PPH versus South Carolina's 14.9 PPH.  If everything holds true to form, expect a Gamecock victory:  27-24 with 56.6% confidence.
Justin: TFG is much more positive on (16) Georgia than they are on (29) South Carolina, although this early in the season we heavily weight home field advantage. Georgia's offense on a neutral field is equivalent to that of Virginia Tech, but their defense still needs some work, ranking on par with East Carolina. For a Steve Spurrier-coached team, South Carolina's offense is surprisingly anemic, tallying a mere 18.1 PPH. That's not a horrible rating, but it's tied with Air Force for the least efficient offense in the TFG top 40. The Gamecocks make up for it on the other side of the ball, claiming the 15th-most efficient defense in Division I-A. This game will come down to how well the Gamecocks can harness home field advantage and ride the momentum to victory. TFG says yes, but only with 55.0% confidence.

Brigham Young Cougars at Air Force Falcons.
GUGS Score: 79.2

Justin: Both BYU and Air Force are coming off excellent seasons that culminated in eerily similar bowl wins: the (21) Cougars demolished Oregon State 44-20, while the (34) Falcons were the brick wall that slowed down Houston's high-powered offense, 47-20. This is a rematch of last year's November game at BYU, but this time the Falcons have the home field advantage. Both teams thrive on defense, allowing 14.2 PPH or less. The bad news for the Falcons is that BYU's offense has a 3.4 PPH advantage; the good news is this is a home game for the Falcons, and TFG gives them the slightest of nods: 30-29, with 56.1% confidence.
Eddie:  Despite the ranking separation between (21) BYU and (40) Air Force, this matchup is surprisingly balanced.  BYU's defense and Air Force's offense are surprisingly consistent, so expect Air Force to put up numbers pretty close to their expected 13.7 PPH.  The deciding factor will be if Air Force can stop BYU's offense.  Traditionally, both struggle against better opposition.  Home field advantage keeps it a little too close for BYU's comfort, but the Cougars should come away with a victory:  28-27 with 62.8% confidence.

Georgia Tech Yellow Jackets at Kansas Jayhawks.
GUGS Score: 75.2

Eddie:  Just looking at the algorithm's output, I should file a bug report -- not because RBA chooses (16) Georgia Tech to win, rather that they should beat (47) Kansas by only one touchdown.  However, the pick seems reasonable when you look at how Georgia Tech has actually played under Paul Johnson.  The Yellow Jacket offense should roll at 21.6 PPH, but their running game tends to result in fewer plays per game than other teams, meaning that they don't exploit that disparity as much as other teams.  Kansas gets the three point bump for playing at home, but their 14.8 PPH worth of offense just isn't going to get it done.  RBA picks the Ramblin' Wreck 35-28 with 68.4% confidence.
Justin: The conventional wisdom about this game is that (20) Georgia Tech is going to stomp all over the (48) Jayhawks. TFG doesn't dispute the Yellow Jackets as the favorite, merely that the home field advantage for Kansas will let them hang around for a while and have a shot at the upset. Kansas will have to establish their defense early and stay close throughout; Tech has the fourth highest-rated offense in D-1A, behind only Florida, Texas, and Alabama. What really hurts the Yellow Jackets is their defense, which maybe one day will be as good as Maryland's. Or -- if they work really hard -- as efficient as UVa's. A lack of defense and home field advantage lets the Jayhawks hang around longer than they should, but ultimately they fall short, 37-35.

North Carolina State at University of Central Florida Knights.
GUGS Score: 69.0

Justin: You're probably thinking "where they hell did GUGS pull this game from?" If so, stop spying on me when I run GUGS and look at the output for the first time. The quick answer is that this is the most interesting game between a team in a BCS conference and one not in a BCS conference (other candidates this week include Michigan-Notre Dame, LA Tech-Texas A&M, and Buffalo-Baylor). This game rates poorly on the "quality" scale, but rates highly on the closeness metric. The (73) Wolfpack visit the (68) Knights in a game that TFG is not too hot on -- UCF is a 2-to-1 favorite -- but between TFG and RBA is pegged to be the fifth-closest game of the week. So if NCSU doesn't manage to make this a bit closer than the 34-29 UCF victory that TFG predicts, blame Eddie.
Eddie:  RBA isn't really a fan of either (68) NC State or (74) UCF, but it expects this one to be tight.  The Wolfpack offense is highly erratic with a 29.4 PPH variance between the strongest and weakest teams.  UCF isn't particularly stable on either side of the ball, wavering 27.5 PPH on both offense and defense.  This leaves us with the Wolfpack defense, which isn't particularly good against anybody, surrendering 16.8 PPH on average.  On a neutral site, this game is a toss-up, but home field advantage tilts the game for the Knights:  31-28 with 54.6% confidence.

Games you probably wanted us to preview:

Penn State Nittany Lions at Alabama Crimson Tide.
GUGS Score: 56.3
This game features of pair of excellent teams, weighing in with the highest quality score of the week. Unfortunately we only give Penn State a 1-in-5 shot of winning this one, so it rates low in the more-heavily-weighted "closeness" factor. Sorry, Penn State fans.

Miami Hurricanes at Ohio State Buckeyes.
GUGS Score: 39.0
This games has the second-highest quality rating, but it's entirely due to Ohio State's ranking. RBA says the Hurricanes have a 2-in-9 shot of pulling off the upset, and that's the optimistic prediction; TFG says 1-in-10.

Thursday, September 9, 2010

Week 2: Thursday and Friday Predictions

With one week of play under our belts, it's time to look forward to today's and tomorrow's games. No big games here, but look for Central Michigan at Temple to go down to the wire. RBA has it as a 60/40 game in Central Michigan's favor, whereas TFG calls it a virtual coin toss with a 0.6% edge going to home team Temple.

Home Visitors Odds Home Visitors Odds
(39)Mississippi St.33 (31)Auburn31 60.3 (26)Mississippi St.28 (38)Auburn24 58.6
(74)Temple33 (51)Central Michigan31 50.3 (73)Temple27 (59)Central Michigan28 60.0
(49)Houston46 (94)UTEP32 86.9 (57)Houston45 (89)UTEP24 72.1
(93)Marshall25 (30)West Virginia32 76.7 (91)Marshall17 (39)West Virginia31 83.3

Week 1: Wrap-up

Week 1 is officially in the books.  It was an exciting weekend with amazing last-minute victories by Boise State, East Carolina, and Jacksonville State.  It's time to take a look back and see how we did with limited information.

Eddie
Accuracy:  33-6 (85%)
Eddie is so smart because:  Boise State beat Virginia Tech by exactly three points, although it was slightly higher scoring than predicted.  Speaking of which, RBA went 65% against the spread last week.
Eddie is a moron because:  Fresno State whipped Cincinnati 28-14.  Although the Bearcats jumped out to a 14 point lead, their offensive line had no answer for a Bulldog defense that sacked Cincinnati QB Zach Collaros eight times.  After a slow start, Fresno State beat Cincinnati in all phases in a game that wasn't as close as the score indicates.
Eddie wants a mulligan on:  Maryland over Navy.  The Middies missed a field goal, fumbled twice at the Maryland one-yard-line, displayed LSU-like clock management before halftime, and were stopped no more than six inches from the goal line on the final play of the game when they could have kicked the field goal to send the game into overtime.
Eddie got lucky on:  East Carolina over Tulsa.  ECU needed a Hail Mary to beat Tulsa.  Enough said.

Justin
Accuracy:  33-6 (85%)
Justin is so smart because:  My computer expected to get 33.037 games correct this week, and had to settle for 33 on the nose. The ones I got wrong were either expected to be close games (BSU/VT, East. Mich./Army) or had such weird stuff happen that no one can or will be able to explain it (Navy gets all butterfingers inside the 3 around Maryland, Fresno State managed to trick Cincy into replacing their defense with the local Girl Scout troop at halftime). I can live with getting those wrong.
Justin is a moron because:  Maybe picking Middle Tennessee over Minnesota wasn't the best idea in the world. I'm having a word with my talent scouts now to see what the hell they were thinking.
Justin wants a mulligan on:  Maryland over Navy. See Eddie's section for a good summary. At least in the football stadium you can't hear the Terps chanting "[eff] you, [opponent] ::clap clap clap-clap-clap::" like you can during basketball season.
Justin got lucky on:  ECU over Tulsa. For those who didn't see it, you should really check out the drive log to see what it was like in the second half. It's like both teams watched the Cincy/Fresno State game and said, "Now Cincinnati ... there's a defense! We can be just like them!"

Wednesday, September 8, 2010

Week 2: Full Rankings -- RBA

I'm having some script issues, so no summary of risers and floppers this week.  As mentioned in the Top 25 summary, I'm working to include FBS games in RBA, meaning that Ole Miss should be pretty much ignored here.  Full rankings after the jump.


Week 2: Top 25 -- RBA

First order of business:  Yes, I know that Ole Miss is way too high.  We have not historically been grabbing FCS games because teams that lose to FCS opponents are generally pretty bad to begin with.  Until the import tool can grab FCS data reliably, we're stuck with Ole Miss in the top 25.  Don't panic -- they'll drop as soon as possible.

Since we're on the subject of anomalies:
  1. Virginia Tech is ranked above Boise State:  From a statistical perspective, Virginia Tech and Boise State played dead even in a game that wasn't settled until a minute left in the game.  The separation between the Hokies and the Broncos has more to do with Texas dropping and Arizona rising than the result of Monday's game.
  2. Major churn in the top 25:  RBA has a higher weight factor on the current season than TFG, so teams have major jumps based upon their performances in the first game.  Since Oregon, Arizona, and Nebraska smoked their week 1 opponents, they had a major jump.  Expect these to smoothen out over time as we observe more games.
  3. Southern Cal's precipitous drop:  The Trojans played poor defense against a vastly inferior Hawaii team, increasing their defensive efficiency by 4.4 PPH. That's going to hurt your ranking.
  4. The Oklahoma love-fest concluded:  Oklahoma has dropped from the top 5 all the way down to 23rd.  That's what happens when you lose 7.6 PPH of offensive efficiency against a bottom 50 team.

Rank+/-TeamWinPctSoSOff PtsDef PtsPace
001--Alabama0.95230.547923.96.579.1
002--Oregon0.94660.531729.28.487.6
003--Ohio St.0.93290.533721.66.980.0
004--TCU0.92550.465723.47.884.9
005--Florida0.91560.557823.88.480.9
006--Penn St.0.90020.539919.77.682.1
007--Nebraska0.89410.531420.68.283.7
008--Virginia Tech0.88790.536420.68.479.2
009--Texas0.87660.513723.610.184.2
010--Arizona0.87480.558622.19.583.2
011--Boise St.0.85920.458725.211.585.1
012--Georgia0.85270.554025.511.980.1
013--Mississippi0.84300.522422.210.781.8
014--LSU0.83790.545519.79.680.7
015--South Carolina0.81580.546518.09.479.0
016--Georgia Tech0.81560.535826.313.882.2
017--Oregon St.0.80960.543421.511.586.2
018--Stanford0.80740.555823.012.384.3
019--Arkansas0.80270.564526.714.582.6
020--Miami (FL)0.80060.550619.110.581.2
021--BYU0.79960.482921.411.785.3
022--Oklahoma St.0.79740.551522.612.585.0
023--Oklahoma0.79740.541520.211.184.4
024--Southern Cal0.78940.535223.313.183.6
025--Iowa0.78670.521616.19.282.2

Expect the top 25 to be fairly fluid until around week 5 when teams stop whipping overmatched non-conference opponents and start playing more familiar teams.

Tuesday, September 7, 2010

Week 2: Full Rankings -- TFG

Biggest jumps: Arizona (+15, 38 to 23); South Carolina (+15, 44 to 29); Iowa State (+15, 85 to 70).

Biggest drops: Southern Miss (-17, 58 to 75); Northern Illinois (-16, 71 to 87); Cincinnati (-14, 19 to 33).

Full rankings after the jump.

Current Rankings -- TFG

Update: as of August 2, 2012, the most recent TFG rankings have been moved to this page.

Week 2: Top 25 -- TFG

This past week saw very little action between top-ranked teams, except for the thriller between Boise State and Virginia Tech. When looking at these rankings, keep in mind that they are by and large still a reflection of last year's season. A mere 11.8% of the statistical weight that goes into these rankings is from the games you just saw, while a full 66.2% comes from the 2009-2010 season. The remainder comes from even older data.

Rank +/- Team WinPct SoS Off Pts Def Pts Pace
001+2 Alabama 0.9422 0.5830 27.4 8.7 79.2
002-1 Florida 0.9275 0.5797 30.7 10.9 77.0
003+1 Ohio St. 0.9272 0.5339 24.3 8.6 81.4
004-2 TCU 0.9102 0.5172 24.3 9.5 83.0
005-- Texas 0.8855 0.5142 27.8 12.2 83.8
006+7 Oregon 0.8847 0.5512 25.3 11.1 90.7
007+1 Boise St. 0.8795 0.4951 26.7 11.9 84.5
008+1 Penn State 0.8788 0.5778 22.7 10.1 80.7
009-2 Virginia Tech 0.8707 0.6519 26.4 12.1 80.3
010-4 Oklahoma 0.8497 0.6212 21.5 10.7 89.3
011+1 Iowa 0.8343 0.6018 21.2 11.0 80.4
012-1 Nebraska 0.8309 0.4498 21.2 11.1 80.4
013+2 LSU 0.8158 0.6409 22.1 12.0 82.3
014-4 USC 0.8118 0.5614 22.6 12.5 84.5
015-1 Texas Tech 0.7723 0.5349 24.6 15.1 88.9
016NA Georgia 0.7643 0.5835 26.2 16.3 79.7
017-1 Utah 0.7572 0.4982 21.4 13.6 84.9
018+2 Mississippi 0.7555 0.6323 21.5 13.6 81.6
019+2 Clemson 0.7448 0.5111 20.8 13.3 81.9
020+2 Georgia Tech 0.7412 0.6616 27.0 17.6 79.0
021-4 BYU 0.7382 0.4596 21.5 14.2 85.2
022-4 Pittsburgh 0.7337 0.5870 23.2 15.4 81.5
023NA Arizona 0.7332 0.5211 20.2 13.5 82.5
024NA Oregon St. 0.7298 0.6649 23.3 15.7 84.4
025NA Arkansas 0.7291 0.6629 25.3 16.8 83.4

New entries: Georgia, Arizona, Oregon State, Arkansas.

Dropped out: Miami-FL, Stanford, Cincinnati, Air Force.

Even after one week the computer polls begin to fall in line with people's expectations. Alabama and Ohio State rise to the top, USC and Texas Tech slip downwards. Georgia gets bonus points for manhandling LA-Lafayette, and Oregon is rewarded mightily for pounding New Mexico. After a few weeks, however, these lopsided wins will start fading into the statistical noise as TFG is reminded that neither LAL nor UNM is a powerhouse this year, and a big win counts for little. Ole Miss is saved by the fact that games against I-AA opponents aren't taken into account; we may have to fix that eventually.

The Pac10 makes a strong start out of the gate, placing 4 teams in the top 25, second only to the 6 teams from the SEC. This early on it's difficult for the computers to pick anything out of the noise, though.