Monday, June 26, 2006
OFFSEASON STAT RELEVANCE WATCH, PART ONE: BIG PASSING STATS VS. VICTORIES
- - - - - - - - - - - - - - -
SMQ got sidetracked this weekend by this now old but amazing article on coach/law school grad/savant/wannabe pirate Mike Leach, which left him thinking, "Yeah, those stodgy old run-first coaches are excessively beholden to old-fashioned "smashmouth" football just because it's traditional and familiar! Where's the innovation? The experimentation? Throw, baby, throw! Swing your sword! Every play! Yeah!" Crush the status quo, man.
This maverick attitude faded by Sunday night, when he looked at Fresno State and got all Woody Hayes about the strong correlation between gaudy passing yardage by Paul Pinegar and FSU losses - Pinegar averaged 373 yards in five defeats, as opposed to 203 yards in eight Bulldog wins - even trotting out an anecdotal "truism" from deep in his conservative, conventional wisdom-loving psyche: "Impressive Passing Stats=Impressive Defeats."
But this statement was made with no empirical backing, and is likely as wrong as any other half-assed perception. So SMQ got to wondering: do Pinegar's stats represent an extreme version of a quantifiable national trend? Or is that idea born of cherrypicking a few surprising instances against a whole of contradictory evidence? What overall correlation do big passing performances have with winning and losing, anyway? What about more run-of-the-mill numbers? Most importantly, how fucking geeky is even thinking about all this?
To find out, SMQ looked at every game by the top 60 passers (by yards per game) last season and put the win-loss result of each game into one of six appropriate yardage-based categories: more than 400 yards, 350-399 yards, 300-349 yards, 250-299 yards, 200-249 yards and less than 200 yards. He then ranked the records for each yardage category by win percentage.
There are some obvious problems to address immediately: this is only the results of 60 quarterbacks, about half the country's starters, and only the top 60 at that - the best statistical players are generally on the best teams, which begs for 'chicken or egg' style skepticism regarding the importance of one primarily individual number amidst many others generated by good players elsewhere on the team - so there is no accounting for the ways the other, less successful half of the nation's quarterbacks' numbers may have affected the results. No consideration whatsoever has been made for the strength of any opponent.
But 60 quarterbacks is a big sample size (661 games, in fact), big enough to expect some kind of correlation between numbers and winning to emerge, if one exists. As always, note that correlation is not causation - the mere establishment of a relationship does not define the nature of the relationship.
Anyway: the win-loss results, by yardage category:
Quarterbacks passing for more than 400 yards: 18-20 (.474)
Quarterbacks passing for 350-399 yards: 28-22 (.560)
Quarterbacks passing for 300-349 yards: 48-36 (.571)
Quarterbacks passing for 250-299 yards: 84-52 (.618)
Quarterbacks passing for 200-249 yards: 106-49 (.684)
Quarterbacks passing for less than 200 yards: 113-85 (.571)
Each category ranking, by win percentage:
1. 200-249 yards: .684
2. 250-299 yards: .618
3. 300-349 yards: .571
4. < 200 yards: .571
5. 350-399 yards: .560
6. > 400 yards: .474
Veddy intaresding: quarterbacks who threw between 200-299 yards in a game were about 26 percent more likely to win said game than guys who passed for more than 300 in any game; passers who threw for under 200 yards were about identically as likely to win as quarterbacks who passed for 300-349 (there were further fractional differences than shown), and more likely to win than those who threw for 350 yards on up. The biggest numbers, in this case, had the smallest correlation to winning - though the corollary didn't hold, as the smallest numbers were only in the middle of the pack rather than on top.
SMQ will hold off any speculation on what conclusions should be drawn from these numbers regarding styles of play and their relationship to winning; this is tricky because throwing for 325 yards means something very different for Texas Tech or Hawaii than it does for, say, Wisconsin (or Fresno State). With the former, the numbers were almost always so high in both wins or losses that they revealed nothing; when all you do is throw, you win by it and lose by it. The Badgers, on the other hand, were undefeated in the half dozen games Jon Stocco kept it under 225, but lost two of three when he passed for more than 300 (he threw for 301 in the bowl win over Auburn). In that case, they run, they win.
So an awful lot depends on the ground game, too, which is why SMQ will return to the stat sheets later this week to perform the same analysis on the nation's top rushers before speculating further on the relevance of any of this.
- - - - -
205 signatures and counting...
Perhaps looking at passing yards in the first 3 quarters?
You can have horrible rushing and passing totals for the first half of a football game, finding yourself down 14 points. Guess what, you're going to finish the game with horrible rushing yards and great passing yards, whether you win or lose. Now, if you're up 14 points at the same point, you're largely going to rely on your running game to burn up clock (unless, like Fresno & other schools, you don't have a running game), and your passing totals will fall off significantly.
That was my point. It seems to me the general notion is "The more yards, the better," and a lot of attention is given to quarterbacks who lead or finish near the lead in passing yards (Dan Marino, for example; Ty Detmer and Andre Ware won Heisman Trophies for this, David Klingler, David Carr, Tim Couch were very high draft picks based largely on astronomical passing totals). Think about it: when a guy throws for 350 yards, you're impressed when you see that stat. That's considered a very good number, certainly better than 250, right? Because more yards should equal more points and therefore more wins.
And, in general, good passing stats do equal wins, as the best passing quarterbacks (as shown here) and the best passing teams are well over .500 as a group. So it's not correct to say, as the ASU fan seems to be implying, that big passing numbers are necessarily the result of losing (certainly nobody is suggesting that losing is the result of big passing numbers). But it might be true to say that this is the case more often than with big rushing totals. Could you take this all the way to the point that rushing yards could somehow be considered "more valuable" than passing yards? I don't know. That assessment is still in the works for later this week.
But I do know that when I looked at team stats, the best passing offenses had pretty much identical overall records as the best rushing offenses. This was also true for the worst teams in each category. So, on a team level, rushing doesn't correlate to more wins.
There are a lot of levels at play here, and I think, as always, balance is the key. Teams that run and throw well are much better than those who only do one aspect well. And one overriding formula can't apply to every game: sometimes teams have to throw more to win, and this frequently is successful. It's just not quite as successful as when they dictate the strategy, and include more runs.
You're just not complex enough to understand that, SMQ.