there wasn't a cluster of teams that really excelled in this area. Sure, most were above 30 minutes of possession per game, but that isn't exactly indicative of ball control football. In my mind time of possession is only really meaningful in games where there is a decisive advantage (e.g. >10 minutes) for one team. Of course, I seem to recall Pittsburgh holding something like a 19-minute edge in time of possession last year when they played Cincinnati. And they still lost.

I have to confess I'm actually not all that interested in determining the ultimate formula for TOP (despite what this thread would have you believe). I'm just reacting to some commonly-held assertions ("Scoring too quickly is bad", etc) and trying to prove or disprove them. Like everyone else here, I think TOP is a bunk measurement and I cringe every time football broadcasters put up their "halftime stats" and marvel over the time of possession. If you want my advice, I would drop it entirely from your series.

- No text -

influenced the results. Per Lafortune Teller's point above, this could be a strong contributor. In other words, is it teams with run-heavy offenses and very good defenses that occupy the majority of the top time of possession spots. I know for the last 10 national champions (I keep referencing these teams because I'm about to put out a series of articles that outlines their common characteristics) they almost all had top 10 defenses coupled with an average run/pass split of 60/40. Could be both that strongly contribute.

- No text -

tend to have strong defenses as much as an excellent defense is requisite to winning a championship. Eight of the last 10 champs ranked in the top 10 in yards per play, yards per game, and points per game. That's damn impressive.

not the cause. Offhand, I can't think of a more pointless stat in any sport that people actually use than TOP. Even total plays seems better to me than TOP. The correlation between total plays and winning should be worse than the correlation between TOP and winning, but that doesn't make TOP a worse metric - the correlation is worse because a team that is winning in the 4th quarter runs the ball more than the losing team, so TOP increases.

Essentially, if you want to know which team won, you could look at TOP, but you'd be better off looking at the score, which turns out to be perfectly correlated with winning. The TOP doesn't provide any information in addition to the score that is not better evaluated with a different stat.

NFL offenses are safer and more methodical. So many teams start taking less chances once they get into field goal range. Once a team gets the lead in the 4th quarter, many coaches will opt to run the football. They don't need style points: they need the win.

College offenses are much different. In many games, one team is clearly much more physically superior than the other team. Lack of quality depth means that the weaker team will probably fade in the 4th quarter. The better team has no interest in putting on the brakes once it takes the lead. Style points matter. If a player is in the running for the Heisman, then he's going to get additional reps to pad his stats.

I'd say there are more "big plays" in college football. Busted coverages, blown assignments, WR A is 5x faster than CB A, et al. As a result, many offenses can score from anywhere on the field. Now throw in fast-paced offenses like Kelly's, or Dick Rod's. Because of that, TOP doesn't really tell the whole story.

Air Force ran one of the fastest paced offenses in the country last year (about two seconds faster per snap than Cincinnati by my approximated tempo measure - read about it in HCTI!), but the Falcons also ranked in the top 10 in TOP. That's because they run the ball so much and "possession time" and "real time" tick off together when you run the ball.

going back to the original contention that when people think TOP, they usually think offense. That the defensive variables are more significant is the extent of the revelation, such as it is. Maybe, you know, people should think defense first. And maybe those concerned about the low TOP for the Bearcats shouldn't automatically assume it's the offense that's the culprit (or that it's even a problem in the first place).

But that's about it. None really explain a whole lot; while the DYPG variable has the highest correlation value, its actual r2 value is about 12.5%. Even combining OYPG and DYPG doesn't really explain much.

I'd wager if you look at turnovers and third down percentages, you'd get a much better read on TOP.

with the post above. BCS champs and contenders exhibit better than 50% TOP.

Why? Because BCS champs and contenders tend to have strong defenses.

in correlation to TOP. (The negative values for defense might be throwing you off; take the absolute value of those for the comparison to offense. For offense, the higher the YPP and YPG the better, but for defense it's the opposite. Hence the negative correlation.)

most of the recent championship teams possessed the ball a significant amount of time. So while having solid TOP numbers may not guarantee success, it does seem to be a common thread amongst elite teams.

"This seems to show us that defense is a stronger determinant of a team's TOP than offense." or the other way around?

The team with the best TOP in the world could be a team with a grind it out offense that never scores and a defense that gives up a touchdown on the first play every time.

The team with the worst TOP in the world could be a team that scores on its first play every time and has a defense that lets teams drive 80 yards on them every time, but stops them at 4th and goal.

These are outlandish examples, but the crux of my argument is that there are much less misleading metrics out there.

I did a quick correlation calc between TOP and Offensive ypp/ypg, and Defensive ypp/ypg.

http://spreadsheets.google.com/ccc?key=0Aqu7YH9xlSk0dDFIb1BnSDQyVmxER2hRR3R0UVJzcUE&...

I ran this for four years' worth of data (2006-09).

It really looks like better defenses (by these measures) are much more strongly correlated to TOP than better offenses. (In the case of Yards Per Game, the difference is pretty stark, I think).

TOP Correlation to Yards Per Play
0.078 Offense YPP
0.120 Defense YPP (+.042 better correlation)

TOP Correlation to Yards Per Game
0.138 Offense YPG
0.354 Defense YPG (+.216 better correlation)

You guys can check my math; I might be all wet on this or I forgot something significant. I threw this together in about 20 minutes, but the results surprised me. I figured it would be a lot closer to even.

This seems to show us that defense is a stronger determinant of a team's TOP than offense.

That said, nine of the last 10 BCS national champions averaged at least 30 minutes of possession per game and 15 of the 20 title game participants fell into the same category. As a whole, the last 10 title winners averaged 31:43 and the 20 participants averaged 31:11. So it seems that it is a common element for teams that achieve at a high level.

I feel like a broken record saying this, but I think people tend to forget that TOP isn't only about whether your offense chews up clock. It's equally dependent (by its very definition) on your defense being able to force punts and turnovers and generally get off the field. But people don't think about it this way. Whenever someone mentions TOP, it's always a conversation about offense, not defense.

I firmly believe that TOP is overrated. What really matters, in my mind, is which team is dictating the tempo of the game, similar to basketball. Kelly clearly prefers a faster paced game, and so people getting flustered about the TOP differential are ignoring that Kelly's teams practice and prepare for games with that a hectic pace. It's part of what gives them an edge.

- No text -

There's been a lot of concern expressed recently regarding Coach Kelly's offense and it's pacing. I believe the argument goes something along the lines of "Kelly's offense doesn't use enough clock up and his defense will wear down both within a game and over the season. The game's different at this level than in the Big East and unless he changes something, he won't succeed." Now, while I think there is some truth in that, I wonder to what degree it's true. I think we need to keep in mind that football is an funky game to consider from the perspective of the game clock. I think that the thought that "lower Time of Possession means a more tired defense" is a pervasive belief, but is only indirectly true. Allow me to elucidate.

Suppose I went to the gym and you asked me what I did. "I was on the elliptical running program W for an hour". Tells you a lot. "I lifted X, Y, and Z at A, B, and C pounds for about an hour" is partially useful: you know i did a few particular exercises at a particular weight. I even did them for an hour, which seems like a good chunk of time. It doesn't tell you a key piece of information: <i>how many reps of each did I do.</i> Unless there's some new weight regime I'm unfamiliar with, knowing somebody was in the gym "for an hour" doesn't really say all that much about the quality of their work out. Likewise, football is a game more similar to lifting than it is the elliptical workout: the number of reps is far, far more important than the duration. So if you will indulge me, I want to take a look at the numbers through that lens.

Here's something interesting to consider. To everyone's chagrin, our Coach Kelly's offense finished dead last in Time of Possession last season, at 25:46.46 minutes per game. Contrast this to the National Champion Alabama, which finished fifth with 33:31.00 minutes per game. So clearly Nick Saban's offense is crafted to chew up more clock, correct? Let's take a deeper look at those numbers to find out. Alabama accumulated 469:14:00 minutes versus 335:04:00 for Cincy. Keep in mind that Alabama played 14 games to Cincy's 13. More importantly: Alabama ran 947 offensive plays to Cincy's 833 (67.6 versus 64 offense plays/game). The math tells us that Kelly's offense was #106 in Time per Offensive play at 24 seconds/offensive play. By comparison, Alabama was #7 on this metric at 29 seconds/offensive play. So really, the "pace" of offensive plays accounted for 320 seconds of TOP and the "lost" three plays accounted for another 87. So the largest contributor to Alabama's greater TOP is that the offense took an extra five seconds between plays. So why can't Kelly do that?

For one thing, Cincy only used 64 plays per game on offense, which was #94 in the country and a full deviation below the mean. What's really amazing then is that Kelly's offense was still #4 in the country in points per game at 38.6. With so few plays used, Kelly's offense was one of the most efficient on a per play basis: at 0.60 points per play, Cincinnati was second in the country only after BSU.
The remainder in the top ten of that category was rounded out by Arkansas, Nevada, Florida, Stanford, TCU, Oregon, Texas, and BYU. Cincinnati performed two deviations above the mean in this category. While Cincinnati was very efficient with the ball on offense, their relatively low number of opportunities meant they had to work quickly when they did have the ball. Had the number of offensive plays dropped in proportion with an increase in five seconds per play, Cincy's already low offensive play total would have dropped to around 52 plays per game, dropping production to only 31.2 ppg.

So it appears that there is credence to the belief that Kelly's offense reduced TOP by design. I would argue that design was beneficial to Cincinnati, since the faster pace yielded nearly a touchdown more per game in offensive scoring. How did this design impact the Cincinnati defense? Given the low TOP, Cincinnati's 71.9 defensive plays per game (#107) should come as little surprise, although that was little more than one deviation below the mean of 67.4. Truly being on the field must have impacted Cincinnati's ability to keep points off the board, right?

As I previously noted, Cincinnati was outstanding in terms of offensive points per play. While they were not quite as strong on the defensive end, they performed quite admirably in this category: 0.32 points against per play, good for #32 in the country (mean: 0.37, deviation: 0.091). The top ten for this stat was Nebraska, Alabama, Penn State, Ohio State, Florida, TCU, Oklahoma, LSU, Iowa, and Virginia Tech, which I think most would agree were very powerful defensive units. I think that performance is even more impressive when you consider the volume of plays against Cincy and also consider the talent of their defense (it wasn't pretty) relative to a Big East offense. Thus, if it is the case that Cincy's defense was tired in games last season, then reduction of that volume is sufficient to go from good to great.

When you look at the difference between offensive vs defensive points per play, Cincinnati stacks up with the top teams in the country: #4 at +0.28. The top ten for that margin in 2009 was Florida, Boise State, TCU, Cincinnati, Alabama, Texas, Virginia Tech, Nebraska, Penn State, and Ohio State. In my mind, that is very impressive.

Conclusions: First and foremost, I think that Kelly needs to be lauded for his work at Cincinnati. On a play-to-play basis, Cincinnati was amongst the best in the country and I think that bodes well for Notre Dame going forward. I think that some of the TOP concerns raised on this board are appropriate: Kelly's defense was exposed to far too many plays last season. If he can get a 50/50 balance between offensive and defensive plays here, I think that will be a harbringer of excellent things to come.