(Points + Rebounds + Steals + ½Assists + ½Blocked Shots – Field Goal Attempts – Turnovers - ½Free Throw Attempts - ½Personal Fouls) / Minutes = Win Score per Minute

Wednesday, May 23, 2007

Frustrations of a Fan

I'm pretty sure this post is coming only from my frustrations at watching the Utah Jazz look like a team of rookies whenever they play in San Antonio. And so the righteous indignation that will come through in this post is probably not going to be properly aimed. But I'm going to let it fly anyway, knowing full well that I am letting my emotions get the best of me.

It's OK.

As I was updating the database this morning with the stats from last night's game, I took a second to look at each of the Jazz players, mentally checking off in my mind how the numbers corresponded with my impression of how they played.

Fisher was terrible....check
Kirilenko, Harpring and Okur were below average....check
Millsap didn't get nearly enough playing time....check
Deron Williams is a stud....check
Boozer played great on offense, terrible on defense....huh?

I was stunned to see that Boozer, according to Win Score, played almost as well as Tim Duncan. How could that be? I watched him fall asleep on the defensive end time after time! I checked Fabricio Oberto's line (the man Boozer was covering most of the night) to be sure, and there it was, 14 points on 6-7 shooting for a man averaging 4-5 points per game through the season. And from my own viewing I knew that at least 5 of those 6 shots were ridiculously easy layups made possible by the lackadaisical defense of one Carlos Boozer.

Now, I've always been a defender of the Wins Produced system, even against those saying that it underrates good defenders and vice versa. There is a very long post on the Wins of Wages Blog that goes over this whole phenomenon of defensive team statistics in great detail. But I guess I had never seen a disparity this egregious with my own eyes before. So I'm concerned.

What is the answer here? There are some possibilities:

1. My eyes deceive me, and Boozer really did contribute as much to a possible win as Tim Duncan did.
2. Win Score and Wins Produced are limited in their ability to evaluate a single game, but the effects balance out over an entire season, which makes for an accurate evaluation of 82+ games.
3. There is a fairly significant flaw in the way Wins Produced evaluates those at the top and bottom of the defensive totem pole.
4. ??

Help me out here, guys. Is there a fourth option I'm not seeing? Have any of you had similar experiences where something didn't add up quite right for you? We all know that there is no perfect metric, so should I even worry about it? It's still the best thing we've got, right? Right?


Anonymous said...

I believe there was a post once on dberri.wordpress.com about how win score could be adjusted to include a +/- opponent points while they were on, but it made little difference

jeremy said...

I've had the same exact problem watching the games and then comparing them to Wins Produced. Understanding how individual defensive performance effects wins is also the one big bugaboo I have when trying to convince fellow fans of the validity of Wins Produced. (They think it can't be done--i.e. you can't account for such a thing in a team game.)

To answer your questions, I think it's a little bit of 1 and 2.

Regarding #1, all of the stats that Boozer recorded mattered. It's my experience time and again that crucial things like rebounding, steals, lack of turnovers, making a high percentage of plain old layups--these things aren't very memorable when you watch on TV. Just because they aren't exciting enough to be memorable doesn't mean they don't add up to a win.

Regarding #2, yes, one game is a very small sample. If there is an effect from exceptionally good or bad defensive effort not recorded in the box score, well, the wildest deviations from WP would be felt here at the single game level.

However, I would say that if you think through the outcome of good individual defense you'll find a high percentage of chances for rebounds and steals. I bet that over the long haul good defense is recorded in a player's ability to net rebounds and steals.