Sunday, July 02, 2006
OFFSEASON STAT RELEVANCE WATCH, PART TWO: BIG RUSHING STATS VS. VICTORIES
- - - - - - - - - - - - - - -
Tuesday, SMQ reported his findings on the game-by-game win-loss records of the nation's top 60 individual passers (by yards per game) last season, which surprisingly - or totally obviously, depending on your outlook - did not correspond in any significant way to the passing total in any given game. Middling numbers, in fact, were more likely than big ones to result in a win, and even the "worst" totals had a higher win percentage than the largest category.
This demonstrated that big, gaudy passing stats in any single game, the kind that bug out eyes and lead to highlights, record books and post-season awards, are often the result of frenzied comeback attempts when some other facet of the game has gone very wrong, attempts that are sometimes successful and sometimes not, rather than part of a blanced, unpredictable attack that wins a greater majority of the time. And that was pretty much the point. But the point is not made without also looking at rushing statistics, to see if the same conclusions hold across the board. So, SMQ also tallied the win-loss totals, by individual games sorted into six yardage categories, of the nation's top 60 rushers from last season.
The win-loss results, by yardage category:
Backs who rushed for more than 250 yards: 6-4 (.60)
Backs who rushed for 200-249 yards: 28-8 (.78)
Backs who rushed for 150-199 yards: 68-23 (.75)
Backs who rushed for 100-149 yards: 115-66 (.64)
Backs who rushed for 50-99 yards: 103-106 (.49)
Backs who rushed for less than 50 yards: 31-85 (.27)
Each category ranking, by win percentage:
1. 200-249 yards: .78
2. 150-199 yards: .75
3. 100-149 yards: .64
4. More than 250 yards: .60
5. 50-99 yards: .49
6. Less than 50 yards: .27
First, sample size: with the top quarterbacks, we were dealing with 661 games, and it's about the same for the top 60 rushers, where we have 643 games. This is probably due to injury, and to the fact that more running backs come off the bench at some point during the season than quarterbacks (only one of the passers in the survey earlier this week, Rudy Carpenter, was a true freshmen, for example, but many of the running backs - probably about a dozen - were first-year guys who didn't start the season).
Anyway, what we see from these numbers is that big rushing numbers are much more conducive to winning - though, it should be noted, there are diminishing returns at the very top of the scale, as with the passing numbers, though not as diminished. Conversely, low rushing totals correlate heavily to losing, as opposed to poor passing stats, which were every bit as likely as big passing numbers to result in a win, and sometimes even moreso.
This is not so earth-shaking, or even interesting. It's perfectly conventional wisdom: winning=running. Not necessarily the other way around, SMQ takes pain to note. Big passing numbers do not lead to losing, but the other way around; in the same vein, being ahead tends to increase the number of handoffs and, subsequently, yards per game. Duh. The point is merely to confirm SMQ's prevously unfounded, anecdotal suspicions that big passing numbers come disproportionately in losses, and relative to big individual rushing totals, that certainly appears to be the case.
What is interesting is that, as noted earlier in the comments of the look at the passing stats, the results of team performance last year does not correlate to the individual results: when SMQ tallied the records of the best rushing and best passing teams last December, the respective results were virtually identical for both the best and the worst teams in each category.
What we can observe from all this is that, over the course of a season, there may be no advantage to either running or throwing well - and only a very, very slight one to combining them well - because almost no team is at its optimal performance every game, and the average bears out the good and the bad. But in micro terms, the good rushing correlates very strongly with - though does not necessarily cause - winning, and low rushing totals go hand-in-hand with losing, in both cases much more so than do good and bad passing numbers.
Whether any of that is of any additional value to anyone's conventional understanding of the game is debatable. At least we know that when Sam Keller throws for 368 against some generous PAC Ten defense this fall, we can guess it's mainly because his own defense is being equally generous.
- - - - -
in general, it is easier to pass block than it is to run block. it takes less size and strength to deflect an opponents momentum away from the QB than it does to force and manipulate your opponent in the direction one desires.
thus, it is my inference that there is a stronger correlation with good rushing (as opposed to good passing) because a skilled offensive line is a requirement for good rushing, whereas a moderate-less skilled line can still get the job done in a heavy pass offense.
now if only offensive line stats were apart of the official stats, we could study this theory empirically.
I think you're right, and I've always maintained that any team that can just line up and run the ball straight ahead is going to do so. Championship teams have badass lines, and usually very experienced ones, which is why OL experience is one of the major variables I'm tinkering with in my formula for preseason ratings right now.
But I'm not extolling the virtues of rushing so much as I am trying to develop a sense of what stats "matter" and which ones don't. They all matter some, and averages at the end of the season show that consistent success in any single major category generally correlates strongly to a winning record. I think the the individual game is more useful, though, because the end-of-year averages don't take into account the micro situations that went into them - that is, if a "normal" team (i.e. one that doesn't throw every down like Texas Tech or Hawaii or run every down like the service academies) goes 7-4, and runs the ball well (for, let's say, around 200 yards) with mediocre pass stats in five of those wins, but runs poorly and throws very well in comeback efforts (maybe 275 or 300 yards) in two of those losses, what is the average going to look like for each of those stat categories? Somewhere in the middle, which is right where the record would indicate.
But the team doesn't perform on the average every time out, and the average doesn't tell us what happened when the team went over or under the number in any given game. This analysis - sort of, since it was only meant to look at individual players, and not team performances - does. And what it says is winning=running (this is not the same as saying running leads to winning; again, it's probably the other way around). There is a correlation.
The most important thing, as always, is that correlation is not causation and, above all, balance is the key. The more aspects a team can do well in any gven game increase its chances of winning, and over, of winning consistently. And remember that big passing also correlated pretty positively to victory - just not as strongly.
Two questions, SMQ:
(1) Are you considering the performance of the opposition in your numbers? By which, I mean, if both teams have a back who rushed for between 200 and 249 yards in a game, the fact that this subset gets a win and a loss counted for it isn't really "fair," so to speak. Admittedly, by doing this you'd probably lose a considerable amount of your sample size (particularly at the lower levels), but at the same time the trends would probably be more pronounced.
(And, I'm well aware, it would take considerably longer to calculate.)
(2) Why aren't you considering team totals, as opposed to individual totals? With QBs, this makes sense--unless you are 'Zona State, it's likely a considerable dropoff to your next QB (think Georgia against Florida, without Shockley). But with the running game, I'd argue that the OL is the biggest x-factor, even moreso than the RB, so deviations among the RBs are less important than those at QB.
Additionally, some teams use a stable of running backs--for example, with USC (since both Bush and White were among the Top 60 RBs), if both backs went for (e.g.) 110 yds in a game, shouldn't the 220 yd total be counted once, as opposed to the 110 yd total counted twice? Wouldn't this be a more accurate representation of whether having a strong running game correlates to your ability to win?
In order to be a good run blocking OL, you just need to be able to come off the line and strike a blow. Since moving forward is more instinctual than moving backwards, and keeping your balance while someone hits you while moving backwards, the kinetics alone prove run blocking is easier.
Add in that in pass blocking, you need to be able to engage your man, b/c you are doing everything possible to keep yourself between the defender and the QB. You have to block that defender for between 3-5 seconds. When run blocking, you can take one step, then dive at the feet of the defensive lineman, cutting him at the LOS. You only have to get him down for a second, b/c the RB has the ball and is gone in an instant. Nebraska's storied OLs of the '90s were experts at cut blocking. You are operatig under the misconception that in order to be able to run block well, you have to be able to physically dominate your opponent, and that simply isn't true. You have to be able to get a stalemate for a split second, and you're done. Normally, running schemes have a double-team taking place at or near the point of attack, so at that one spot, you need to be able to physically move someone, but when you consider all that requires is an extra back to come over and help the lineman at that gap, you're not talking about something difficult.
There is a reason so many youth and HS teams use run-oriented offenses; it is easier to teach and more basic to execute.
Pass blocking is also an individual intensive endeavor, while run blocking is a team-intensive one, which is why it can be easier to teach run blocking. If you screw up while run blocking, your teammates can pick up the slack. If you screw up pass blocking, it's a sack.
All of those are good points, and I have to admit this was not a very extensively-planned experiment - especially in the rushing area. It wasn't meant to be a catch-all correlation exercise, just a quick exerpt of some numbers to check what I consider to be the conventional wisdom about statistics (more is always better). This was designed really to address just one specific point.
That point was to look at passing stats, and whether, as I suspected, big, gaudy passing numbers came hand-in-hand with losses on a regular basis. This guess was substantiated to a large degree: when you see a big passing performance on the scrolling ticker, check the circumstances before handing this guy the Golden Arm Award.
As far as rushing's correlation to winning, I included it only to show that what I was demonstrating with passing numbers wasn't also the case on the other side, not to prove anything overall about handing off. It's not a true evaluation of how closely winning hews to rushing, and vice versa, but it does show that, much more so than passing, a big individual rushing performance will predictably result in a win a pretty large percentage of the time. It doesn't show anything beyond that, as far as overall team success - I did the averages for good and bad rushing teams over an entire season, of course, but that doesn't necessarily give a full picture for the reasons I mentioned - and it wasn't designed to do that.
I am planning to come up with a way to track these trends week-to-week over the upcoming season, and maybe we can get a better overview then (and, most importantly, one that doesn't require me to go through the stat sheets of every game all at once).
So consider this, in the spirit of blogging in general, an imperfect, haphazard venture that answers the question it asked, and brings up about five more it's not useful for answering at all. Se la vie.
i disagree with everything you say, except for that LT's are a premium in the NFL. i would rebuke you point by point, but i am tired from the finals i just took and i need to nap because i turn 21 at midnight..and i'm getting atomically thrown.
since we both speak with such authority on the matter, and as pointed out above, there is 0 empirical data to analyze. i'm gonna go with our individual credentials as the tie breaker.
i was a 2-year starting center for a Texas 5A HS Football team that ran a multi. formation/Spread offense. for those not familiar with texas football: Texas 5A > the MAC > Sun-Belt > Big 8
I look forward to your point-by-point rebuttal. I'm from Texas, so I'm very familiar with Texas HS football, in all of its various classifications.
As for my credentials, I coach youth football, so I actually have to TEACH kids how to play the game. Trust me, it is a lot easier to teach run blocking than it is pass blocking. I can take a kid with average talent and average feet, and with some work, make him into a great run blocker. However, if he has average feet, he isn't going to be a great pass blocker, no matter how much I work with him. It's simply a matter of kinetics. People are more comfortable moving forward and delivering a blow than they are moving backwards, and taking a blow.
It is easier to teach a kid to be the aggressor, and charge out and get a man, than it is to teach him to move back, and let the player come to him.
I am curious at to what HS you went to, though.
I wouldn't brag too much about being the center; I know of too many coaches who hide their worst offensive lineman at center, figuring the doubleteam help from the guards will cover him up. This is why Bud Wilkinson invented the 3-4 defense, to take advantage of that weakness.