by Bill Connelly

There are basically four levels at which to look at college football statistics: the year level, the game level, the possession level, and the play-by-play level.

The year level is most often just looked at in terms of wins and losses. Team A went from 7-5 in Year A to 5-7 in Year B and 3-9 in Year C; consequently, Coach A was fired. There isn't much, statistically, in this.

The game level is extremely common. Team A is averaging 37.3 points per game. Quarterback B is averaging 298.7 passing yards per game. You can rank teams and players with this, but much of the underlying story is still left untold.

Football Outsiders readers know that the possession level is Brian Fremeau's specialty. His Fremeau Efficiency Index derives ratings and prediction from (non-garbage time) possession data.

That leaves the play-by-play level. It is the reason Varsity Numbers exists. As the college football regular season comes to a close, now is the time to start looking at what play-by-play data can tell you about the season you have just witnessed.

To do that, we will first revisit a couple of concepts discussed earlier in the life of this column. The first is S&P, which you can find defined in the Football Outsiders glossary. It is the OPS of college football statistics, combining efficiency (success rates) and explosiveness (Points Per Play).

The second is the "+" concept, a primitive attempt at a college-level look at DVOA-type comparisons discussed here. The formulas behind the "+" concept have been strengthened since the writing of that column (instead of looking at game totals and comparing them to opponents' per-game averages, it actually looks at per-play totals, eliminating the general per-game problem of small sample sizes), but the idea remains the same: Figure out a way to factor a team's strength of schedule into statistics to truly find the best offense, defense, and possibly overall team. You read "+" ratings the same way you would read OPS+ or something similar in baseball: 100 is exactly average, under 100 is bad, and over 100 is good.

(If you want to look at this as a DVOA figure of sorts, just subtract 100 and view it as a percentage. So a "+" figure of 122 could also be viewed as +22%. A "+" figure of 90 would be -10%. The Overall "+" figure below is the same way, only the base is 200 instead of 100.)

One caveat with this post: Play-by-play entry without a parser like the one we use in the NFL (and like the one we're in the middle of building for NCAA data) is time-consuming, and only the 2008 games involving BCS teams have been entered to date. True "+" ratings cannot be calculated until all games have been entered. So why are we discussing this topic today? Because it is conference championship weekend, and there is a lot of interesting data to share regarding conference play.

BCS Conference Breakdowns

So let us dive into the world of BCS conference data. The tables below show you conference breakdowns of the following statistics:

Rushing S&P+ (offense and defense): A comparison of a team's rushing output to the output expected based on the number of rushes against the team's opponents.

Passing S&P+ (offense and defense): A comparison of a team's passing output to the output expected based on the number of passes against the team's opponents.

Close-Game S&P+ (offense and defense): This is an overall S&P+ figure for all plays run while a given game was "close." What makes a game "close" has been tweaked from previous Varsity Numbers columns as well. Previously, "close" simply signified a scoring margin within less than 17 points. This has been altered in the following ways:

• First quarter: If the scoring margin is within 24 points or less, the game is "close."
• Second quarter: If the scoring margin is within 21 points or less, the game is "close."
• Third and fourth quarters: -f the scoring margin is within 16 points or less (i.e., two possessions), the game is "close."

Overall, this is a relatively minor change, but it does signify that an early 21-point lead is much "closer" than a 21-point lead in the second half.

Being that a team's performance while a game is still more or less winnable is what is most important (and when a team gives up a ton of yards and points when they're up 49-7, it really is not that important), the Close-Game "+" numbers are used for the next measure as well.

Overall "+": This is the measure by which the teams in these tables are ranked. It is simply Offensive Close-Game S&P+ plus Defensive Close-Game S&P+. In this category, 200 signifies average, not 100.

Projected Record: In theory, if you know how many EqPts per game Team A averages rushing and passing, and if you know how far Team B usually holds opponents above or below their season averages, then you can come up with a figure that represents Team A's likely output against Team B, and vice versa. Throw in a home-field adjustment, and you can project likely results and therefore a team's likely record based on their (conference) season averages.

In other words, if every team played at its average level in every game (which will obviously never happen), you can project what their record likely would have been. Differences between actual and projected records could be explained by far too many variables to count -- special teams breakdowns, good or bad luck, good or bad coaching, or maybe the simple fact that some teams get much better or worse as the season progresses.

So with all of these explanations and caveats in mind, here are the major "+" figures, rankings, and projected conference records for the six BCS conferences. Each category below does not work out to a perfect 100.0 average because each team ran a different number of plays and therefore has a different denominator.

Thoughts and observations:

• Best Rushing Offense: 1) Missouri, 2) Texas Tech.
• Best Rushing Defense: 1) Texas, 2) Missouri.
• Best Passing Offense: 1) Oklahoma, 2) Oklahoma State.
• Best Passing Defense: 1) Texas, 2) Oklahoma.
• Considering OU also had a tougher non-conference slate than Texas, this does appear to give you a case that, looking at the entire resume top to bottom, OU had as good or better a case for the BCS (and Big 12) title game than Texas. That certainly will not serve as any sort of consolation for Texas fans (45-35!!!!!!).
• Running back Derrick Washington has been a wonderful weapon for Missouri this year, and if they have any chance of beating Oklahoma this weekend, he's going to need a huge day.
• Speaking of Missouri, you will also see that they have the No. 3 close-game defense in the conference. Aside from Oklahoma, nobody has given up more junk points and yards than the Tigers, who spent nine of 12 games comfortably ahead and in a soft, preventative zone. That said, anybody who watched the Missouri-Kansas game knows they have a mediocre-at-best pass defense, and that means they likely will not hold down Oklahoma's offense any better than anybody else has over the last month.
• When you are attempting to pick a winner in the Big 12 North for 2009, if you find yourself tempted by Colorado, make sure to take in just how far they have to advance just to have an average offense.

Thoughts and observations:

• Best Rushing Offense: 1) Florida, 2) nobody close.
• Best Rushing Defense: 1) Tennessee, 2) Alabama.
• Best Passing Offense: 1) Florida, 2) Georgia.
• Best Passing Defense: 1) Alabama, 2) Florida.
• Florida has by far the most dominant offense within their conference of any BCS team. In a conference full of slug-it-out, slogging offenses, to say Florida's stands out is a severe understatement
• If you want to further view the "+" ratings in the DVOA prism, consider Tennessee the Philadelphia Eagles of these ratings. They are inexplicably high. Their strong performances over the last two weeks of the season (after Phil Fulmer's retirement/resignation) aided their numbers greatly, but this does suggest that Tennessee doesn't have too far to go to bounce back into the Top 25.
• The Volunteers' high rankings and low projected record also suggests that they had just about the roughest possible schedule. They played the Nos. 1 and 2 teams in the conference at home (and they were not close enough to get a projected win), and they played the Nos. 5, 7, 9 and 10 teams on the road. Flip the home/road slate, and they probably pick up at least one more win (against Auburn).
• Kudos to first-year Mississippi coach Houston Nutt and Rebel quarterback Jevan Snead (not to mention underrated wide receiver Mike Wallace) for putting together a strong passing game and overall offense. It was good enough to sneak Ole Miss into the No. 3 slot in the rankings.

Thoughts and observations:

• Best Rushing Offense: 1) Florida State, 2) Georgia Tech.
• Best Rushing Defense: 1) Virginia Tech, 2) Wake Forest.
• Best Passing Offense: 1) North Carolina, 2) N.C. State.
• Best Passing Defense: 1) Clemson, 2) Florida State.
• How crazy a year was it in the ACC? The No. 4 and 5 teams in the conference, in terms of overall performance, made it to the ACC title game. Of course, that is what you can expect in a conference where 11 of 12 teams finished between 3-5 and 5-3.
• Further craziness: Not a single ACC team was above average (i.e., over 100) in all four major categories -- rushing offense, rushing defense, passing offense, and passing defense. Boston College was closest, missing by just 0.6 points (a 99.4 in passing defense).
• In their first two games against FBS competition, North Carolina State managed a combined 400 yards and nine points. By the end of the season, they had the No. 3 close-game offense in the ACC. The future is potentially bright with quarterback Russell Wilson at the helm.

Thoughts and observations:

• Best Rushing Offense: 1) Penn State, 2) Iowa.
• Best Rushing Defense: 1) Iowa, 2) Penn State.
• Best Passing Offense: 1) Ohio State, 2) Penn State.
• Best Passing Defense: 1) Ohio State, 2) Penn State.
• The numbers say this should have been a three-way race for the conference title instead of just Penn State vs. Ohio State (with Michigan State attempting to make some noise and not quite succeeding). Iowa's three conference losses were by a combined 11 points. Add to that a 1-point non-conference loss to Pitt, and you have one of the unluckiest teams in the country. It's doubtful that sophomore quarterback Ricky Stanzi strikes fear in the hearts of opposing fans, but he has led one of the most deceptively strong offenses in the country over the last couple of months. Granted, A) they bolstered their numbers with a 55-0 thrashing of Minnesota in the Metrodome's college curtain call a couple weeks ago, and B) passing the ball is going to be a lot easier when you've got a strong runner like Shonn Greene in the backfield.
• Meanwhile, Michigan State's numbers simply weren't all that great. They had two main strengths in '08: 1) Javon Ringer (1,590 rushing yards, 21 touchdowns), and 2) good karma after last season's impressive "six losses by a touchdown or less" display. This year they were 2-1 in "one possession or less" games, and their overall numbers were mostly hurt by debacles against Ohio State (a 45-7 loss) and Penn State (49-18).
• If not for a putrid passing game, Michigan would have just been slightly below-average. But their inability to pass left them with no sustainable offense, and the projected records suggest they were lucky to win one conference game, much less two.

Thoughts and observations:

• Best Rushing Offense: 1) Oregon, 2) USC.
• Best Rushing Defense: 1) Arizona State, 2) California.
• Best Passing Offense: 1) USC, 2) Oregon State.
• Best Passing Defense: 1) USC, 2) UCLA.
• Three conference games remain: USC/UCLA, Cal/Washington and Arizona/Arizona State.
• As would be expected, USC was the No. 1 overall team by a significant margin. Oregon, however, was almost as clearly the No. 2 team. In the end, the Ducks had an unfortunate schedule; playing the Nos. 1 and 3 teams in the conference on the road prevented them from making a serious Rose Bowl run. An underrated aspect of Oregon's team was their solid, experienced defensive line; they started three seniors this year. They will be hurt in 2009 by the loss of defensive end Nick Reed and defensive tackle Ra'Shon Harris.
• With three one-possession wins (against USC, Arizona State, and Arizona), Oregon State was the anti-Iowa. They should not have been anywhere near contention for the Rose Bowl, though allowing 694 yards and 65 points to Oregon last week did skew their overall numbers.

Thoughts and observations:

• Best Rushing Offense: 1) Pittsburgh, 2) West Virginia.
• Best Rushing Defense: 1) Cincinnati, 2) South Florida.
• Best Passing Offense: 1) Rutgers, 2) Cincinnati.
• Best Passing Defense: 1) Cincinnati, 2) Connecticut.
• These numbers do not include Rutgers' Thursday night shellacking of Louisville, but even before last night it was pretty clear just how significant a midseason turnaround Greg Schiano had engineered at Rutgers. The Scarlet Knights began the season 1-5 before ripping off six consecutive wins. The two-game gap between their projected and actual records can most likely be explained by the drastic difference in product between early-season Rutgers and late-season Rutgers.
• Rutgers quarterback Mike Teel looked horrendous in the season-opening blowout losses to Fresno State and North Carolina in particular, but he managed to finish the season with a quite respectable 148.5 passer rating and 23 touchdowns. In their final four conference games, Rutgers averaged 50.3 points, and while Thursday night's game will not be enough to catch Cincinnati -- who was quite clearly the Big East's best team, top to bottom -- it should solidify their overall No. 2 ranking in the conference.
• These numbers also paint an interesting picture about West Virginia. While their offensive struggles have been discussed in this space before, it appears that what held them back the most was their inability to stop the run.
• Six of the eight conference teams fall between 190 and 210 on the Overall "+" number, which, along with the fact that nobody scored above 100 in all four major categories, suggests that Big East parity is at or near the same level as ACC Parity.

Summary

In the end, data at all levels tell a different story, even when you get down to the possession- or play-level data. Just look at North Carolina; the possession-level FEI data thinks very highly of North Carolina's work. At the play-by-play level? Not so much.

One thing play-level analysis misses that possession-level data obviously does not, is the importance of third-down conversions and passing downs situations in keeping drives alive or killing them. That is something that we will begin to explore next week with second- and third-level points.

Posted by: Bill Connelly on 05 Dec 2008

13 comments, Last at 24 Jul 2009, 7:47am by Hammer