|
|
Post by mariner42 on Feb 19, 2009 16:43:29 GMT -6
I read this online, it's up at Spreadattack's blog, too. I think it's provocative to say the least and would love to see how the idea transfers to football. Thoughts? Can we measure a player's selfishness/team first play by quantitative analysis in the same way that he does or is football's complexity and innate nature too much? Link: www.nytimes.com/2009/02/15/magazine/15Battier-t.html?_r=1&em |
|
|
|
Post by dal9000 on Feb 19, 2009 18:21:50 GMT -6
Thoughts? Can we measure a player's selfishness/team first play by quantitative analysis in the same way that he does or is football's complexity and innate nature too much?
Of course we can't -- because quantitative analysis doesn't deal with intangible things like "selfishness" or indeed "team-first play." You may object that they have tangible EFFECTS, but I really don't think they do. Not in football, anyway.
What I mean by this is -- OK, try to imagine that you want to create the perfect football stat. The One True Holy Grail of Statistics that will, in a single number, tell us a player's quality or lack thereof, no matter what position he plays. What would that stat be? If it were up to me, it'd be percentage of assignments completed.
Why? Because, when you strip football down to absolute bare bones, there are only two outcomes for a player on a given snap: he can either 1) complete his assignment successfully or 2) fail to complete it. Right? Right. (And, yes, yes, you couldn't POSSIBLY measure this objectively even if the statisticians somehow knew what the assignments were for every player on every snap. But this is a thought experiment, not a practical proposal.)
'kay, so GETTING BACK ON TRACK: imagine that, armed with this One True Stat, this numerical Excalibur, you set out to discover whether a player is selfish or team-oriented. You're going to have a problem: all you know, and all that MATTERS, is whether a player fails in his assignments more or less than his peers do. And if a player fails more than his teammates, is that because he's selfish or is it just because he's bad? There's no way to tell -- and even if you could tell, it wouldn't matter. The end result would be the same.
So I think that being a selfish player, or a team-first player, is irrelevant to how good you are at the actual act of playing football. Are there kids who are selfish? Yep! Are there kids who are gung-ho? Yep! But those are personality traits, not football skills, and you conflate 'em at your peril.
|
|
|
|
Post by dubber on Feb 23, 2009 13:55:44 GMT -6
|
|
|
|
Post by calicoachh on Feb 23, 2009 15:58:00 GMT -6
great article
|
|
|
|
Post by spreadattack on Feb 23, 2009 19:27:32 GMT -6
As an idea for non-stat positions, especially OLine or on defense, the new Lions HC described something he has done in the past: www.slate.com/id/2092863/When reviewing game film, Schwartz uses a simple grading system: He gives a plus (positive impact), a minus (negative impact), or a zero (no impact) to each player on each play. "You take those and then you can push them into an equation," he says. "You basically have an 11-variable equation and the result is yards gained. Over the season, over 1,000 plays, you can isolate a variable." Schwartz hopes to use his data to make personnel decisions: If a minus play by a defensive lineman costs the team on average more yards than a minus play by a linebacker, then perhaps linemen should be more of a priority in the draft or free agency. --------- That obviously can be useful without plugging them into excel and running a regression (google "regression analysis"), and also to put on the bulletin board after the game. |
|
|
|
Post by brophy on Feb 24, 2009 12:04:06 GMT -6
i have been beating my head over this nice article, but I'm struggling to see how a guy like Battier's contributions would carry over to football, but isn't this kind of apples and oranges?
In baseball / basketball / hockey the action is rather organic, whereas football it is primarily static, on a play-by-play basis and you either did EXACTLY as your assignment or you didn't. Therefore, you could find the guys who executed their assignment. So, if a player can execute his assignment on a given play, he grades out high, regardless of stats.......doesn't that 'unquantifiable' become quantifiable when we grade?
|
|
|
|
Post by John Knight on Feb 24, 2009 12:13:13 GMT -6
I have seen some grading systems that make you scratch you head and say, if everyone graded out 80% how did we get beat 35-7?
|
|
|
|
Post by spreadattack on Feb 24, 2009 12:24:13 GMT -6
I have seen some grading systems that make you scratch you head and say, if everyone graded out 80% how did we get beat 35-7? That's why Schwartz wanted to take the next step and compare those grades to the ultimate outcomes, hoping then that you can identify positions/players/etc that are more or less important for winning/scoring/scoring def., which then leads you to make better decisions next time, in theory. To be honest, as great as the Lewis/Battier piece is, it is itself kind of short on stats. The plus/minus isn't a perfect statistic, and a lot of the evidence about what a good defender Battier is is anecdotal ("look at his feet" "look at his hands" "oh there he forced the guy to his left," etc.) I love the article, but it's all still ongoing. |
|
|
|
Post by brophy on Feb 24, 2009 12:29:48 GMT -6
I have seen some grading systems that make you scratch you head and say, if everyone graded out 80% how did we get beat 35-7? if that is correct (kids follow the plan to a 't') then the plan (play call) was what lost the game. You could grade out at 75% and miss 3 tackles..........with 3 runs going for scores, right? I dunno. The grades would be a foundation (the starting point) towards putting all the pieces together. Where do you fit a guy like Battier on a football team? Where do you hide this 'winner' utility player? The first page had me thinking of a guy like this;  But then they further clarified what made the unspectacular Battier spectacular. Football is offense/defense/special teams....then 11 sub positions, with exclusive assignments. Players aren't as multifaceted as they are in other sports. I'm still trying to understand it, because its 'new paradigm' formula doesn't seem like any 'new' approach for football |
|
|
|
Post by dubber on Feb 24, 2009 12:48:47 GMT -6
i have been beating my head over this nice article, but I'm struggling to see how a guy like Battier's contributions would carry over to football, but isn't this kind of apples and oranges? In baseball / basketball / hockey the action is rather organic, whereas football it is primarily static, on a play-by-play basis and you either did EXACTLY as your assignment or you didn't. Therefore, you could find the guys who executed their assignment. So, if a player can execute his assignment on a given play, he grades out high, regardless of stats.......doesn't that 'unquantifiable' become quantifiable when we grade? I think the article Chris just linked to examines this point. It seems that the "hidden stats" are harder to see in football, because of assignments. You may not get a blocking stat, but that FB has a job to do, and either he did it, or he didn't, and we can grade him as such. However, I suppose something of a "hidden stats" in football may be the WR who, whenever he runs a fade, draws double coverage, giving the other recievers a better chance to get open. If I were to run the fade, however, I WOULD NOT get double covered. However, I executed my assignment, and would grade out the same as, say, Randy Moss. However, Moss' VALUE to the overall goal of the team, of the offense, of that particular fade route, is HIGHER, because he draws double coverage, and I do not. I suppose that could be the type of quanitification/qualification we are looking for...... |
|
|
|
Post by Yash on Feb 24, 2009 13:31:56 GMT -6
If James Harrison does exactly like his assignment says he is supposed to, he blitzes kurt warner in the superbowl and warner hits boldin for a TD, instead he drops into a zone picks off a pass and runs it for a TD putting another trophy in the steelers trophy case. do you grade him down for not doing his assignment on the play?
thanks for the correction Brophy
|
|
|
|
Post by brophy on Feb 24, 2009 13:40:56 GMT -6
If Jerome does it,  I think the refs would stop the play. j/k - good point by dubber & Yash |
|
|
|
Post by dubber on Feb 24, 2009 14:44:52 GMT -6
If James Harrison does exactly like his assignment says he is supposed to, he blitzes kurt warner in the superbowl and warner hits boldin for a TD, instead he drops into a zone picks off a pass and runs it for a TD putting another trophy in the steelers trophy case. do you grade him down for not doing his assignment on the play? thanks for the correction Brophy There's the point......you can quantify assignments, but can you qualify them. Quantitatively, I did my assignment right 50 out 55 times. Qualitatively, on the times I did what I suppose to do, the ball went away from me 45/55 times (maybe I play behind a good 3 technique). The times it came my way, I got my fit, but the 3 tech. made the play 4 times for me. I also missed 5 tackles, and dropped a pick. Definite lack of quality. If we are going purely by pluses and minuses, I grade out at 90%. But I suck. Meanwhile, my SS counterpart grades out at 80%, yet he garnered 8 tackles, 2 picks, and 4 break-ups..... OK, so the answer seems to be "don't just make it plus or minus, make it a scale of 1-10". Even if we could decide on a scale (Rushing TD=8, Tackle for a loss=4, etc.), sometimes garnering points is NOT IN THE BEST INTEREST OF THE TEAM (like the Battier piece says about selfish point guards looking for assists). In football, this could be the Cornerback who intercepts the ball on 4th and 20 and is promptly tackled, he would have done much better to simlply knock the pass down. Or, with the defense out of timeouts, on 3rd down, and up by 1 point, the running back who gains a first down and then breaks it into the endzone. After the PAT, that team now has an 8 point lead, but will give the ball back to the other team, who may drive the field, score a TD, and get a 2 point conversion. It is unlikely, but the odds are better than if the running back had feel down after getting the first (which would have shrink the opposing team's chances from slim to basically none, as most teams can down the football without fumbling it back to the other team). And again, as teh article brings up, am I getting stats because I am good, or because my teammates are? Personally, I can tell you I made all-conference linebacker BECAUSE of the two DT's playing in front of me. This is not lip service. One was All-State, and the other had simular statistics, etc. The one thing I knew is THEY BOTH DEMANDED double teams. That means I never had an OT trying to block me all season. My stats were good, but it had little to do with me. |
|
|
|
Post by dubber on Feb 25, 2009 11:35:56 GMT -6
|
|
|
|
Post by dubber on Feb 25, 2009 11:42:15 GMT -6
Here it is: Right. A minimum of 3.5 play probably has an average of around 5-7 yards per play, which is a heck of a play (you will break a few). That's very different than the original post though. One further point on the assignment grading thing. Notice that Jim Schwartz did not say that you get a "plus" for doing your assignment. He said you get a plus for "having a positive effect on the play." This is different. That way he learns more about what is, and what is not important. Grading assignments is also important, but different. For example, let's say you're grading your outside linebacker. His assignment is to rush off the edge on the defense's left. The offense runs to the defense's right. Tackle for 1 yard gain. He did his assignment; he had no effect on the play, no points. Next play they run at him, he misses the tackle, negative point. Following play, they run at him, the offense fails to account for him, but the runner takes him on -- he makes the tackle for a 5 yard gain. He did his assignment, but was not aggressive enough. He made the tackle. Maybe a no point or even negative point play. In this way, you will get an accurate picture of what players and what positions are having a positive or negative effect on the play. Take a defensive tackle; they double team him, he takes up both blockers (can't combo up) and the LB makes the tackle. Positive effect. Next play they double team him but combo to the LB; no point or maybe negative. Finally, same hypothetical as dubber said with Randy Moss: they double him, he runs a good route and forces the corner and safety to deal with him, they hit the TE in the seam. Again, positive for Randy Moss. Next play they double him but he loafs and the safety can play both TE and him, he gets a negative or a no point. Does that make more sense? In that case you won't get the 85% for the whole game. It's a different metric than "did you do your assignment," but did you matter to the outcome of the play (positively or negatively). Yes, sometimes they can do everything right but the play will go the other way, but over time you should get some interesting results, especially when you run those numbers against yards/points/wins/etc. The other question I have is how to quantify a player's impact as a decoy. This would seem to me to be harder to grade. I've heard it somewhere (probably from a TV announcing idiot) that certain teams have invented unique coverages for Randy Moss and Steve Smith. l suppose if Randy runs a fade on a draw play, and takes two guys with him, then he impacted the play. However, on 2nd and short, an obvious PAPing situation, does Randy get credit when the offense rips off a 12 yard run because the Strong Safety was out of position trying to keep Moss from running a skinny post off a play fake? I mean, he might not even block well, but he still influences the play, simply because of what he MIGHT do. |
|
|
|
Post by spreadattack on Feb 25, 2009 11:51:14 GMT -6
As an example, in the big heydey Patriots/Colts games the Pats quite frequently used an interesting coverage. It was like 0 / 1/2, or inbetween Cover 2 man and cover 1. They would man up everyone, bring five or six, and the F/S, at or after the snap would roll to Marvin Harrison's side.
This let the corner play very aggressive, and they baited Manning into a few interceptions when he tried to get it off hot to Harrison. The clip below is an example, though not very good quality.
The other side was all man, but with pressure they had a confusion advantage. (At the end of the clip from the behind the QB shot you can see the F/S run over to that side while the other side is straight man.)
|
|
|
|
Post by mariner42 on Feb 25, 2009 13:49:33 GMT -6
I think Dubber makes good points regarding the idea of decoys and the difficulty in grading them, although Spread's post is a good example of how you could demonstrate a positive influence going for a negative result. As a fly team, this could help us determine who's giving the most effective fakes and what plays we really got someone to bite on consistently.
I also like the idea of most plays having a neutral result for Schwartz's metric. Maybe throughout the game the OLB gets a -2, while the DE gets a + 6 because he's more disciplined/aggressive, but they can each grade out the same in terms of sticking to their assignments.
I'm thinking one way this can be helpful is creating a diagram with the Schwartz metric for the last game/entire season on a position by position basis, showing WHERE on the defense your positives and negatives are. Could lead to a good understanding of how teams might try to affect/attack you, as well as help shape who you want to put in what position to succeed (i.e. blitzing certain LB positions more to cover up poor reading, covering a weak run game by bringing down your safety, who's +9 for the season, into the box, etc).
Spread: I was wondering if you could PM a 'recommended reading' type list for this summer when I get more time on my hands? I'd love to know what you're reading for learning.
|
|
