Right you are... Case in point, Temple is #10 in time of possession. Ooooo... Temple.
As usual, your grasp of statistics leads to some interesting points, most of which are valid. What you ignore in this analysis, however, is that TOP does have an effect on the stamina of a defense over the course of the game. Defenses can get worn down quicker than offenses, making it easier to hold on the to the ball AND score, it's a snowball effect.
I'm surprised that you've overlooked this rather simple concept.
Good point. Now we just need someone to start keeping track of points per possession for all the teams, or at least all the teams in the top 25, or in the Big Ten. I nominate everyone but me. I'd like to see it, but I don't want to do the work.
Quote:
"Defenses can get worn down quicker than offenses, making it easier to hold on the to the ball AND score, it's a snowball effect."
Yet look at the decrease in scoring as the game progresses.
I think TOP is a reflection of other things, which are NOT the same between games/teams/etc... So I would say TOP does not effect the defenses stamina... not being able to stop an offense effectively makes the D tired. Which could be a great Offense, or a crappy D, or penalties, etc... So TOP might be handy to say one team defense is on the field longer, but that does shed specific data on why? Were they behind and passing alot? Did they have an ACC offense?
I fully concur on the importance of exhaustion. When you keep your offense on the field, by using a long drive with a significant mix of plays, this wears the other defense down. The old adage about "when you pass, 3 things can happen, and 2 are bad," is particularly true in the 4th quarter. It makes sense when you have the lead to limit your passes. This keeps your offense on the field longer (when you have 4 to 8 yard gains through running rather than 20 plus gains through passing and scoring.) It also opens up more opportunities for your receivers. If the defense is looking for run, run, run, run, and all of a sudden, you throw down the field to MM or AA or to the sideline to SB, the receivers are more apt to be open because the other team is determined to stop the run. I would partially attribute Hart's long run near the end of the game a couple of weeks ago to exhaustion on the part of the defense. This also keeps your defense fresh, and with a limited number of plays (because the other team is 3 and out), your defense will play extremely hard because they know, in terms of personal stats and performance, that they have 3 downs to contribute . . . after that, the ball is back in the hands of the offense.
I suppose I can agree that TOP is an irrelevant statistic. But I will say, I believe long drives that march down the field, that culminate in a touchdown, that last a significant number of plays, both physically and emotionally wear down the other team. While the existence of a good passing game to balance the run game is critical to open up the run, I think that a ratio of 2 runs to 1, 1.5 passes is great.
Maybe the only downside I can think of is that if we face a stiff oponent (think OSU), one could ask if our lack of experience in playing from behind or from even will hurt. While this may be true in some cases, particularly with a newbie qb, I think that another factor trumps the lack of experience. That factor is the element of surprise. When we played ND, they didn't know what hit them. They were totally unprepared for the pass. As much as the team talks about playing "one game at a time," I think it is only wise to keep some of your weapons and offensive schemes under wraps until truly needed. This year, I am hoping that Carr unveils some new wrinkles against OSU when needed, and does the same thing against USC or TN or FL or whoever in the NC game. I was struck while at the Rose Bowl 3 years ago? that Pete Carroll changed a bunch of things between the end of the season and the bowl game, leaving Michigan completely befuddled.
Morphing into different topics yet again, that bowl game gives me hope. I recall how SC's defensive front four manhandled our Off. line. They didn't really need to blitz most of the time, and when they did, it destroyed us. I am convinced that our defense can do the same thing, and conversely, that with our OL and with Hart and our receiving corps, we won't be nearly as susceptible to the same
I heartily agree that football needs to start figuring out what the real/useful/essential stats are, and to start measuring them. On the other hand, I wonder if declaring jihad on TOP might constitute throwing the baby out with the bathwater.
You say that “the reason Michigan is good isn't because they possess the ball for relatively large amounts of time but rather those crushing [run offense & defense] units.” But, as other commenters have already noted, TOP might have an additional kernel of truth in that those units become more crushing because Michigan has an edge in TOP. That is, the longer we have the ball the more the opponents’ d-line is getting gassed and ours is staying fresh, adding to the efficacy of our run offense and defense. It might further be good indicator for overall likelihood of defense breakdowns, particularly with an aggressive defensive scheme where (ostensibly) there is less chance for brainfart because of a tired player.
My two cents.
JP Coltrane--
I disagree. A screen pass over the middle for 5 yards on first down takes about 40 seconds off the clock. That same screen pass for 5 yards on 3rd and 3 takes 20 seconds off the clock (because the clock will stop on first down). That same pass dropped by the receiver will take 5 seconds off the clock. It's the same play, so the defense gets equally tired on all of them, but anywhere from 5 to 40 seconds tick off the clock. Time of possession tells us nothing.
Number of plays from scrimmage might be useful in estimating defensive fatigue, but nobody ever talks about that as a measure. Time of possession? On the screen several times a game. Number of plays? Almost never.
As Brian says, the fact that time of possession is biased in favor of the "old-fashioned" running game has much to do with the fact that it is over-rated as a stat.
Time of possession is a statistical EFFECT of other causes, that in itself AFFECTS other aspects of the game.
Begin with the premise that the offense has the advantage over the defense because they know the play. Team runs ball a lot. Clock runs. Offense on field, opponents defense on field. Offense must SCORE for TOP statistic to have any BENEFICIAL meaning. If M scores 7 in an 8-minute drive, then opponent scores 7 in a 2-minute drive, the score is tied. However, the TOP statistic is in M's favor. What does it mean in a tie game?
In a game of physical exersion and stamina, the team that is more fresh in the second half has a better chance of winning the game. Tired teams do not perform well. If the defense is gassed because it is on the field for 40 minutes, then it is THEORETICALLY easier for the TOP-favored team to score.
TOP is caused by how the offense operates, AND how quickly the defense shuts down the opponent. Having possession longer affects the stamina of the opponent's defense, whose job it is to prevent you from scoring.
There is nothing better in a close football game than having a 100-yard rusher lining-up to run a play against a defensive line and LB crew that is panting and gassing. Unless that defense has some man-child on the field, or other form of bionic man, the team with more TOP will most likly benefit in the end...and the score at the end is all that matters (see PSU-M 2005 game).
Therefore, TOP can be a significant statistic.
I haven’t heard the blathering announcers in the recent games because I watched them in a bar so I don’t know exactly what caused this outburst. On that note, did I hear Spielman call Henne Chad Hennington like he was morphing into the Jets QB?
I think it is accurate to say that TOP correlates well with successful football teams without being the cause of the success. The correlation makes it an important stat but just holding onto the ball does not itself equal victory. My guess of the most 2 highly correlative and causative stats regarding success are run defense (name me a great team that can’t stop the run) and turn over differential. The ability to run is also important but as the Pats and the West Coast offense have demonstrated, there’s more than one way to skin that cat. Those three stats all factor into TOP. I suspect stopping the pass correlates less with success because passing is harder. Look at ND. They are OK at stopping the run but horrible at pass defense. Yet, only the good teams can abuse them for that. Those good teams can really lay the wood to ND though, creating these 9-3 seasons with some really bad losses.
However, there is meaning in UM’s TOP. Once they are up big they can effectively eat the clock. The great criticism of the run and shoot was those teams had a hard time holding leads because they couldn’t kill the clock. Just plunging into the line is only effective if you get to keep doing it over and over because you get first downs. If you can’t run with a lead then your other option is to keep scoring which may be harder and riskier.
Clearly, points per possession is an important stat. If you score efficiently then you’ll be hard to beat. I’m not sure the basketball analogy is that good because basketball is all about a quality shot. If you take them then the odds favor you. Unless you play a press or fast break then most of your possession will come out of the half court offense making most possessions the same. Football isn’t quite the same. All possession are not created equally regarding time and field position. With so many fewer possession those “bad or good possessions” greatly affect the numbers. More importantly, in end game situations winning football teams give up scoring potential to kill the clock while winning basketball teams are given free scoring chances by fouling. That makes score per possession more correlative with success in basketball than football.
In one respect, you can deny your opponent the ball. If you are winning by 2 scores your opponent needs 2 possessions to win. You may not need any more points if you can deny them time to have those 2 possessions.
Re: "exhaustion." Even in that case, a better metric is plays, which erases much of the variance included in the excellent screen example offered earlier. I think there's something there, but there's obviously a more accurate measure available.
If the point is that TOP is a pretty useless statistic, I agree. There can be any number of reasons why a team may have a high TOP. The reason Michigan is #1 is a lot different than why Temple is #10. I've heard football analysts praise TOP as a wonderful thing, the higher the better, for the various reasons stated above. And that, just isn't true.
But Yost--
Team A runs the ball up the middle for 3, 3, 3 and 71 yards, taking 2 minutes to score.
Team B throws passes to the sidelines for 10, 0, 0, 30, 0, 0, 15, 0, 0 and 25 yards, taking 1:40 to score.
Which defense is more tired at this point?
Why not use number of plays instead of time of possession to measure fatigue?
At the same rate I have to argue against yardage defense stats being kind of useless. Your team's stats are skewed in the same way your TOP stats are skewed...get the lead, and pass pass pass to try and even the game up. That's why scoring defense is much more useful than yards. Who cares if a team gives up 400+ yards a game if they give up less than 10 points a game? I'll take the 119th ranked rush and pass defense if that same defense is the #1 scoring defense (alright, that will never happen, but you get my point).
Re: On that note, did I hear Spielman call Henne Chad Hennington like he was morphing into the Jets QB?
I don't think he spent any time mentioning the name of any Michigan player because he was so fixated on that "controversial" TD catch.
Geez, he probably forgot he's not playing for tOSU anymore. Get over it. Leave the whining to LC.
I heard Spielman say "Hennington" as well. He was just summarizing the qb-wr combination on that particular touchdown. Efficient former Detroit Lion, that Spielman.
One thing that hasn't been mentioned: time of possession also measures your offensive discipline. Even big-play offenses can have a tough time grinding out play after play - defenses aren't the only units that can lose focus being on the field for 10+ plays.
Another note: no stat should be taken by itself. TOP goes well with third down efficiency. If all I saw were the TOP and third down conversion numbers from a game, I could probably predict the winner. Win one of the two, you got a 50%+ chance of winning; win both and you're almost guaranteed. Win both plus turnovers and if you don't win the refs must have been from the Big Ten.
The stat that really counts, to me, is the number of plays. While there are (usually) an equal number of possession, what you do with those possessions is variable. I think number of plays run is a very good indicator of success in the first half of games (and probably overall, although it may become distorted by teams that go conservative at the end). If it's halftime and Team A has run 35 plays to Team B's 25, it's a good bet that Team A is ahead. At the least, it's almost certain that it has gained more first downs.
"I don't understand why John L. Smith didn't throw his flag and have the play reviewed," Speilman rhetorically asked, over and over through out that game.
"But Mr. Speilman," I calmly responded to my television after each question, not at all clenching my fists or muttering about the incompetence of the commentators, "coach's flags are in the NFL. In the Big Ten, the coaches don't have flags to throw on the field."
Actually, the coaches DO have challenges this year. It's not all booth review. Whether the implement of challenge is a flag, or an extended middle finger, I don't know. Probably a flag, though.
It's not that DEFENSES get tired faster than offenses, making TOP important. It's that DEFENSIVE LINEMENT get tired faster than OFFENSIVE LINEMEN, which mean that:
a) Less sacks/runs for losses
b) An extra few yards per run
c) Easier to get first downs and control the ball, making comebacks difficult/impossible.
The downside is that in order to get that advantage you have to run more, which makes it more difficult to score a bunch. So while SCORES decrease quite a bit, you can
a) More effectively shorten a game in which you have a lead
b) rest up your own defense to allow them to go 100% (which, Hermann, doesn't help if they're passing and you aren't blitzing)
c) force them to bring 8 in the box to help out the defensive line leaving WRs open (ta-da)!
Basically, TOP makes your run game more effective with all else (i.e. the ability of the defense to guess run) equal. For a team like Michigan, that's really important.
Now, there's of course serious correlations between "good running teams" and "TOP teams", which would make quantification difficult, but there you go.
c)
Alton - you make a good point that the number of plays run by that passing team matters. I only SUGGEST that TOP is not a worthless statistic as Brian believes. Since I believe TOP AFFECTS other facets of the game (i.e., opponents' defensive effectiveness), it can be a valuable indicator. For example, if Team A controls TOP 40 minutes to 20 minutes, but is consistently outscored, TOP could help prove that the offense is not converting all its chances. But, then again you don't need TOP to prove that, you could use red zone %. It seems to me, though, that in a world of choices I'd rather have more TOP than my opponent.
Justin has a point - on run plays, the offensive line has the initiative and the defensive line is forced to react. On pass plays it's the opposite. Probably 95% of offensive linemen prefer run blocking to pass blocking, because it's easier, more fun, and less tiring.
I think most of you are missing the point. What Brian is saying is that winning the TOP battle does not make a team good; rather, it is a byproduct of the fact that the team IS good.
Like the first poster pointed out, Temple is #10 in time of possession, yet they're probably the worst team in D-1. The reason is that they give up quick, easy TD drives about 8 times a game, and then have slow drives on offense that result in no points.
To simplify, it doesn't matter if you run 50 more plays than the other team. If they're scoring on every possession, they're still going to win. It's all about efficiency.
This is really fascinating. I think I'll have to review it an almost infintite number of times to make any sense of it (as I do with any sort of scientific or mathematical set of propositions), yet I think there is some real sense to it to all. To my (very) unscientific mind, it would make sense to divide the points scored by TOP, or at least devise a statistic to reflect this reality. Obviously, the limit to this would be the length of the field, and the first-down rules (how plausible is it to possess the ball from any starting point for 7 -8 minutes on a possession and not score?). On GameTracker, I find myself instinctually looking at turnovers, net offensive yards and TOP (not necessarily in that order) as reflections of the current score. Secondly I look at 3rd down conversions, or at least I think I do. I tend to think, as of the reading of this post, that the best indicator of TOP as a factor is the relationship of scoring with respect to TOP. Theoretically, a team can score on a long pass on 1st and 10 and eat up 15 seconds. But can it do this for the WHOLE game, or enough of the game to make a statistically signicant difference? I don't think so. Thanks to all here for a real nutcracker.
That should read "statistically significant" difference. Oi, the travails of digital communication. Written in stone as it were.
TOP in itself is not a meaningful statistic, but it is a lazy one: it combines many factors together to create a stat that correlates with overall success more than any one other statistic. that is why it has any value: TOP is heavily weighted toward defensive prowess (except in rare situations, like temple, where the defense is so bad that opposing teams score extremely quickly) but also indirectly adds in turnovers created, turnovers lost, and effectiveness of the run game and short passing game. the offensive portions of the game are also weighted toward the "exhaustion" factor. so, in sum, while it may be variable per game, when averaged over the long haul, it does have some meaning as a composite statistic.
i do agree with the suggestion that a points per possession could be more meaningful...i'd also add an opponent's points per possession as well as a gauge of defense and find an algorithm to combine the two so that you have a total view of the team--that would be the ultimate statistic.
Defenses can get worn down quicker than offenses, making it easier to hold on the to the ball AND score, it's a snowball effect."
Yet look at the decrease in scoring as the game progresses.
Fatman | 10.12.06 - 12:08 pm | #
A better picture would be to look at the yards per carry at the end, as a worn down defense is usually easier to run the ball on ,and so the clock.
Holding onto the ball doesn't necessarily win games but leading games makes Michigan more likely to hold onto the ball. Got it. But then wouldn't saying "I hate TOP" be tantamount to saying "I hate Michigan winning"? TOP is your friend, Brian.
At the same rate I have to argue against yardage defense stats being kind of useless. Your team's stats are skewed in the same way your TOP stats are skewed...get the lead, and pass pass pass to try and even the game up. That's why scoring defense is much more useful than yards. Who cares if a team gives up 400+ yards a game if they give up less than 10 points a game? I'll take the 119th ranked rush and pass defense if that same defense is the #1 scoring defense (alright, that will never happen, but you get my point).
Chris | 10.12.06 - 12:51 pm | #
Chris brings up a very good point. As everyone is aware, Notre Dame has one of the worst defenses in the country. They are near the bottom in terms of yards given up. However, they are 5-1, and in terms of "points allowed" by the defense, they are much better than average.
For example, Notre Dame's defense allowed GT to only score 10 pts (the same GT team that scored 89 pts in their last three games, including 38 on the road vs. VT), and only 3 pts to PSU (not including 14 garbage pts). Of course, on the flip side, teams that either a.) throw well (UM), or b.) have large, oversized fullbacks (MSU), have shredded our defense, and that is where they have struggled. But Purdue put up 403 total yards before their 4th Q. garbage drive, and yet only managed 14 points.
ND is one of those weird defenses that "bends" but doesn't "break" - us ND fans have been dealing with such behavior since 1990. Maybe that's why we have been "Returning to Glory since 1993."
I am so cynical...how desperately do I need a great season and another National Championship - 1988 is too long ago.
To my (very) unscientific mind, it would make sense to divide the points scored by TOP, or at least
devise a statistic to reflect this reality. [...] Thanks to all here for a real nutcracker.
diogenes
As long as you've got your lamp fired up - not being that big a football fan, I have no idea if someone
already calculates this, but...
It seems to me that one candidate for such a statistic would be LeadTime, defined as Sum(Time of Lead X
Size of Lead). That is, if Team A scores a TD with a minute left in the game and wins 7-0, their LeadTime
would be 7. If Team B scores a TD as time runs out in the first half and wins 7-0, their LeadTime would be
210. If Team C falls behind 3-0 with eleven minutes left in the first quarter but pulls out a 6-3 win by
scoring as time expires, their LeadTime would be -168. And so on...
Clearly this statistic wouldn't be perfect for ranking teams, as the losing team can have a larger LeadTime
than the winner, but it would certainly separate teams by style - teams that get big leads and successfully
sit on them would have much better LeadTimes than teams that mount furious come-from-behind rallies and
then tack on late scores on turnovers. Which style is ultimately more successful? I have my suspicions, but
I don't know - at least until Diogenes does all the work for us.
But LeadTime would clearly be more useful than plain TOP in separating the Temples from the Michigans...
Ugh, nice linebreaks.
Let me try again:
To my (very) unscientific mind, it would make sense to divide the points scored by TOP, or at least devise a statistic to reflect this reality. [...] Thanks to all here for a real nutcracker.
diogenes
As long as you've got your lamp fired up - not being that big a football fan, I have no idea if someone already calculates this, but...
It seems to me that one candidate for such a statistic would be LeadTime, defined as Sum(Time of Lead X Size of Lead). That is, if Team A scores a TD with a minute left in the game and wins 7-0, their LeadTime would be 7. If Team B scores a TD as time runs out in the first half and wins 7-0, their LeadTime would be 210. If Team C falls behind 3-0 with eleven minutes left in the first quarter but pulls out a 6-3 win by scoring as time expires, their LeadTime would be -168. And so on...
Clearly this statistic wouldn't be perfect for ranking teams, as the losing team can have a larger LeadTime than the winner, but it would certainly separate teams by style - teams that get big leads and successfully sit on them would have much better LeadTimes than teams that mount furious come-from-behind rallies and then tack on late scores on turnovers. Which style is ultimately more successful? I have my suspicions, but I don't know - at least until Diogenes does all the work for us.
But LeadTime would clearly be more useful than plain TOP in separating the Temples from the Michigans...
I think it's the old rectangle-square arguement....All good teams are TOP temas, but NOT all TOP teams are good teams.....would anyone expect Texas, OSU, USC, OU, Miami etc to be terrible in TOP? nope. only possible exceptions may be Surrier at UF and Tx Tech. They are pass, pass pass, score type teams that are clearly the exception, not the rule.
I think a fair stat for judging the most effecient offenses would be something like Yards per possion or points per possion. BUT, one needs to remove field position from the eqn. obviously, NDs 1st 7pts really came from the effectiveness of their D to return an INT to the 4. should the Offense get ALL the credit for staring at the 4 and scoring 7 pts? Nope. I may crunch some numbers & pass them to Mgoblog to post, if so desired.
Commenting by HaloScan
