August 19, 2010

Warriors Stumble On... 35 Wins

You may remember an article that created a bit of a stir in the Warriors blogosphere a month or so ago... Jeremy Britton, a Dubs fan and Wages of Wins contributor, pointed out that the new-look Warriors projected to be a 50-win team by WP48. Actually, the exact projection he came to was 56.9 wins, a number so outlandish that he didn't even dare mention it in the text of the piece.

In case you couldn't tell, we ain't buying this 50-win talk. Frankly, we ain't buying WP48, either. Kevin Love, a guy on a 15-win team, is the fourth-most effective player in the league? Troy Murphy, a massive detriment, a blight upon the earth, is the twelfth-most effective player in the league? Like fun, Wagers. As Dave Berri's evasve snippiness in the comments section attests, this isn't a system you can really vouch for. All you can do is claim that it's infallible, eyes shut, fingers in your ears, shouting to the heavens. We Worriers aren't statisticians, but we know an echo chamber when we see one.

So we're going to project this team's potential ourselves, using two different tools: the Win Percentage figures over at Basketball Prospectus and the Win Shares figures at Basketball Reference. This'll be some quick and dirty figgerin', and the results won't be the gospel by any means, but we'd take them a lot more seriously than the WP48 results. Why? Every key Warrior, except for Dorell Wright, presided over worse team defense when on the floor than their individual numbers suggest. These two metrics attempt to account for that; WP48 doesn't.

So. First, let's estimate minutes totals for the '09-'10 Warriors.

PG: Stephen Curry (36), Charlie Bell (12)
SG: Monta Ellis (36), Reggie Williams (12)
SF: Dorell Wright (30), Reggie Williams (18)
PF: David Lee (30), Brandan Wright (18)
C: Andris Biedrins (30), David Lee (6), Dan Gadzuric (12)

This is an oversimplification, of course, and a pretty Pollyannaish one in several ways. We're not likely to see every rotation member play 82 games, for one thing, and Biedrins will have to get back on track in a big way to play a half hour a night. We're also assuming that Nellie will largely avoid smallball, a dangerous assumption given recent history. You can quibble with some of the numbers here... Jeremy Lin may usurp Charlie Bell, Udoh should eventually grab some minutes, and obviously the Warriors could sign a guy or two who'd figure into the picture. But this, roughly speaking, is the plan.

Next, we convert each dude's playing time into a percentage. A guy who plays all 48 minutes would take up a total of 20% of the available player-minutes. Accordingly, guys who play 36 minutes, as we project Curry, Monta and Lee to do, take up 15% of the available player-minutes. Going down the list:

Stephen Curry: 15%
Monta Ellis: 15%
David Lee: 15%
Andris Biedrins: 12.5%
Dorell Wright: 12.5%
Reggie Williams: 12.5%
Brandan Wright: 7.5%
Charlie Bell: 5%
Dan Gadzuric: 5%

Now, if we weight each player's '09-'10 results by their playing time, we ought to be able to get a reasonable estimate of the win total each system would predict. (We'll use Brandan Wright's '08-'09 data for simplicity's sake.)

Stephen Curry: .519
Monta Ellis: .484
David Lee: .635
Andris Biedrins: .525
Dorell Wright: .502
Reggie Williams: .508
Brandan Wright: .579
Charlie Bell: .334
Dan Gadzuric: .408
Weighted Total Win %: .518 x 82 = 42.5 Predicted Wins

By our rough estimate, Win Percentage is somewhat bullish on the new-look Warriors roster, predicting a record a bit over .500 with perfect health. The metric sees David Lee as an excellent player (though it isn't as starry-eyed as WP48), it regards five of the other top six players as above-average contributors, and even Monta comes off okay here. Win % thinks Dan Gadzuric and Charlie Bell stink, but they're hardly alone on that one.

Stephen Curry: 0.077
Monta Ellis: 0.023
David Lee: 0.163
Andris Biedrins: 0.101
Dorell Wright: 0.137
Reggie Williams: 0.104
Brandan Wright: 0.131
Charlie Bell: 0.060
Dan Gadzuric: 0.077
Weighted Total WS/48 = 0.494 x 82 = 40.5 Predicted Wins

Win Shares doesn't echo Win Percentage on a granular level -- it likes Monta and Curry much less, and the Wright Brothers much more -- but it comes to a similar overall conclusion. Enough talent to be competitive when healthy, not enough elite talent to really approach "good".

While we're having fun, let's irresponsibly throw another stat on the pile.

Stephen Curry: +4.50
Monta Ellis: -2.95
David Lee: -2.45
Andris Biedrins: -3.76
Dorell Wright: +6.20
Reggie Williams: -7.50
Brandan Wright: -1.22
Charlie Bell: -1.21
Dan Gadzuric: +1.67
Weighted Total APM = -0.836; Pythagorean Estimate: 38.6 Wins

APM isn't really meant to be used this way... it does not necessarily follow that, say, David Lee's plus-minus numbers as a Warrior forward will mirror his results as a Knick center. This is the noisiest and least reliable data set. Still and all, its results are interesting, and more than a little troubling: an APM-based approach doesn't even see a freakishly healthy Warriors team cracking 40 wins. Curry and Dorell both rate as big, big pluses, but every other key guy rates as a negative, generally due to wretched defensive numbers.

By any accounting that acknowledges how a team's defense was affected by the presence of a given player, these Warriors do not rate as having 50-win potential, nor anywhere particularly close to it. They look like a .500 team. And while you may believe that guys like Curry, Monta and Biedrins will play better this season than last, you've also gotta believe that the Warriors' top seven guys will not all play significant minutes in all 82 games. There are more reasons to tug that 41-win estimate down than up, given how poor the team's injury replacements project to be, and given the fact that Don Nelson looks ever more likely to retain the ill-fitting title of "head coach" to start the season.

This is a team with a decent starting lineup and a very bad bench, a team with very little defensive talent, a team with a coach with very little interest. The Dubs should crack the 30-win mark for the first time since Baron left, but barring a major trade or a Nellie firing or both, they probably won't hit 40. And there's no real reason to expect them to compete for a playoff spot, either. If we assume the Lakers, Mavs, Spurs, Thunder, Blazers and Nuggets are all playoff-bound, the Warriors will have to finish at least second in a scrum that includes the Rockets, Hornets, Jazz, Suns, Grizzlies and newly Griffin-ed Clippers. David Lee has never even led a team into the Eastern Conference postseason, and while Stephen Curry is very good, he'll be competing with Chris Paul, Deron Williams and Steve Nash for a playoff spot... it's hard to argue that he gives the Warriors an edge.

Trading Anthony Randolph for David Lee is pretty much the epitome of a "win now" move. The Warriors don't seem at all likely to win now. What was the point of that, again?


Anonymous said...

Very enlightening and beneficial to someone whose been out of the circuit for a long time.

- Kris

Anonymous said...

this post is very usefull thx!

Anonymous said...

Great information! I’ve been looking for something like this for a while now. Thanks!