Monday, July 13, 2009

Some Early Prediction Thoughts

all stats taken from

I'm researching and exploring some methods to utilize in the hopes of making a more objective, more accurate prediction for the 2009-10 season with the goal of getting within, let's say, ten games of the Hawks' final record. A small step but one which would constitute an almost 30% improvement for me over last year's prediction. It's all about setting attainable goals. That last sentence is surely a petard on which I'll be hoisted.

This first table looks at Win Shares, cataloging each player's rate of Win Shares per 1000 minutes for his career, the 2008-09 season, and while playing for the Atlanta Hawks.

WS/1000 Min

NameCareer08-09w/ ATL

One can pretty easily take those rates, make an educated guess about each player's 2009-10 rate, assign minutes for 2009-10 season, and calculate a rough estimate of the team's projected wins. Something like this...


Last year's 47-35 team (46-36 Pythagorean record) earned 45.1 Win Shares.

The reasoning behind that Jeff Teague projection will be the subject of a future post which I'll preview here only by saying that point guards drafted outside of the lottery over the last six seasons, even point guards who have gone on to be above average players, have pretty much all been below average players as rookies. Collectively, they've been well below average.

Though Win Shares is a useful tool for a rough exercise such as this (not least of which because one needn't make hundreds of calculations) it doesn't give you much of an idea about the shape of a player's contributions. Thus, as a complement to Win Shares I've also looked at each player's Offensive Rating (ORtg) and Defensive Rating (DRtg) for the 2008-09 season and their careers.


Two things to notice in the above table:
  1. Lots of Hawks exceeded their career averages (last season inclusive) during the 2008-09 season. I think it's fair to expect the young players (Horford, Williams, Smith) to continue outperform their career averages even if they might not improve upon or even repeat their 2008-09 numbers. I think it's fair to expect the older players (Bibby, Crawford, Evans, Pachulia) to perform closer to their career averages in 2009-10. As for Joe Johnson, he's posted (in chronological order, 2005-2009) Offensive Ratings of 109, 109, 109, 110 and Defensive Ratings of 114, 111, 112, 110 in his four seasons in Atlanta. I suspect he'll post similar rates next season.
  2. Notice the similarity between Flip Murray's 2008-09 and Jamal Crawford's career. I still don't think that Crawford is likely to provide an upgrade over what Flip Murray produced last season.
There are a couple of extra steps to projecting a record from individual ratings. I estimated each players minutes per game (eMPG) for 2009-10, multiplied their ORtg and DRtg by (eMPG/48), added up the results, divided both sums by 5 to get estimated Offensive and Defensive Efficiencies, and then used those to figure pythagorean wins. The estimates below resulted in an Offensive Efficiency of 109.4, a Defensive Efficiency of 108.2, and 44.3 wins.


Now, as to whether these objective measures confirm my suspicions that the Hawks have merely re-constituted last season's team (assuming they re-sign Marvin Williams) or whether this post just confirms that, before the fact, I can only conceive of avenues of inquiry which confirm my suspicions, I'm essentially neutral. What say you?


JMar said...

Poor Zaza. He's only 25 and getting lumped into the older crowd. It's tough being in the NBA.

jrauch said...

There's a good number of Eastern Conference NBA teams that actively sought to improve their rosters this past year, and teams that clearly improved throughout the course of last year that should be more concerning in 2009-2010.

Reconstituting last year's Hawks this year won't be getting it done in trying to take that step to the next level (ie: Winning a game against Cleveland in the playoffs).

Bronnt said...

I'm not completely sold on win shares. One of its big weaknesses is that it only takes into account offensive and defensive rating, and only in terms of point per possession.

There's no concession to things like turnovers and rebounding, which create extra possessions for the offense and defense, respectively. So it's not perfect as a player evaluating tool.

Xavier said...

Out of curiosity was this same thing done before last season? What was the prediction? And was this done for the Magic.

Jack Bender said...

Interesting analysis. I am looking for more impact from Crawford than you seem to be. I think he will be playing more crunch time minutes in the 4th quarter than Murray did in 2009. He gives Woodson more line up options. You could play Bibby, Crawford, JJ, Marvin, and Horford together--or go Smoove in for Bibby.

Bret LaGree said...


I'm not sold on any of the single number, cumulative evaluation metrics but I'm not capable (due to both time and knowledge constraints) of creating my own player projection system and, as far as I know, there's no such open-source tool for basketball players. This should be considered rough estimation.


Not be me. I'm sure had it been done, it would have predicted fewer than 40 wins. Both Hollinger and Kevin Pelton had the Hawks winning 30 some games before the season. I believe both made a caveat about their systems underrating Joe Johnson and thus possibly the team but it was unexpectedly good shooting from Bibby, Marvin Williams, and Flip Murray that made the team exceed projections.


I'm afraid you might be setting yourself up for disappointment with regard to Crawford. He's a league average player replacing a guy who was, quite surprisingly, a league average player last season. 85% of Crawford's shots last year were jump shots (by comparison 79% of Joe Johnson's attempts were jumpers) and he's not an especially good jump shooter (40% on two-point jumpers last year, 34.7 3PTFG% for his career. Whatever help he gives by lightening the load for Johnson and Bibby will likely be given back by limiting the opportunities for Horford and Marvin Williams to improve their production.

rbubp said...

Crawford, like drafting Teague and Acie Law and getting Flip Murray, is yet another case of hoping a player will improve in certain directions rather than trading for or selecting players based on what they are already capable of. It worked out with Murray, but didn't wit Acie, so we'll see about Teague and Crawford.

The other thing Bret did not mention is that Crawford is not a particularly good three-point shooter either, not to mention below-average in every other category: assists, rebounds, steals, blocks.

I think the problem here is that Woodson is fixated on combo guards and players who can create their own shot. With more of the the latter, he has less responsibility to design actual plays and put players in positions on the floor to be most successful.
In effect, he don't gotta coach as much.

Just roll the ball on out there, Woody.