2011-2012 HDR Player Ratings Available

02/03/2013

6 Comments

 
I just finished the HDR player ratings for the 2011-2012 NBA seasons, and they are available under the HDR Player Ratings tab.  


A few changes were made to the formula between the 2012-2013 ratings, and the just published 2011-2012 ratings.  The two big ones include:


1.) Counting a missed shot as -.87 possessions as opposed to -.75.  By using the value of .75, I was essentially giving the shooter all the credit for a possible offensive rebound, which of course is silly.  It should be evenly split between the shooter and rebounder, hence the new value of .87.  


2.) Incorporating a defensive adjustment.  This was accomplished using the Defensive Rating metric, which while not perfect, does a decent job of accounting for defense.  I'm working on standardizing the values with respect to positions, but until then, we can do with my current method.


That's all for now.  I'm working on a few projects, including a test for how accurate HDR is as a predictor of team success.  If you have any questions about where a certain player ranked, and why, feel free to hit the comments.  
 


Comments

JonCBK
02/06/2013 1:55pm

Hmm, I'm just seeing the top ten. Your methodology is really resulting in some very high numbers for a few players and then by the time you get to the end of the top ten in each position the rating has dropped dramatically. Your rating is certain supporting the concept that the NBA is a star driven league and that a good star seems to be worth nearly as much as an entire team of average players.

Reply
HoopDon
02/06/2013 4:40pm

Yeah, I just did the numbers for players who played 25 min per game, and played in at least 30 games in 2012. I want to get a better handle on what we are trying to measure, before I crank out the numbers for every player in the league.

You are correct. A true superstar is worth quite a few "good" players.

Reply
Shrinidhi
02/06/2013 3:19pm

Nice work! This addresses some of the key areas where Wins Produced seems to be on slightly shaky ground. That said, subjectively altering the metric as we see fit may to what we think makes sense may not be the best course of action.

For example, why did you choose -.87 as the value of a missed shot? Why not -.82 or -.93?

Also, why did you choose Dean Oliver's Defensive Rating over any other defensive metric (I believe Synergy Sports has one)?

I like where you're going, but I'm afraid that if we just subjectively alter Wins Produced as see fit, we may end up going backwards rather than forwards. It'll take (a lot) more work, but objectively justifying the changes we make will go a long way to increasing the usefulness and credibility of the metric.

Reply
HoopDon
02/06/2013 4:54pm

The values aren't subjective.

Wins Produced counts missed shots as -1, or a turnover. On a possession level that is problematic, because that indicates every missed shot is a rebound for the defense. That's not true, as historically, a offense has a 25% chance of rebounding their missed shot.

As such, I used to think a field goal attempt should be worth -.75, to account for the offensive rebounding chance. However, that would in a sense, be giving the shooter all the credit for potential offensive rebounds. It should be half and half, so we add half of .25 to .75, for a rough value of .87.

Defensive Rating is the only defensive metric that is available in bulk. I think Synergy does a better job of measuring defense, but its data only covers 1-3 years, and is extremely difficult to use when analyzing numerous players at once.

Reply
Shrinidhi
02/06/2013 8:53pm

Yes, I understand, but what I was asking is -- why half and half? We've established that we can't give no credit to the shooter and that we can't give all of it to the shooter. But that doesn't mean we should split it 50-50. That's the subjective part. I could easily argue that the offensive rebounder deserves more credit, if not most of it.

And DRtg is a fine metric. It should work well. Looking forward to seeing the projects you're working on!

HoopDon
02/07/2013 7:20am

Its really the same issue with assisted shots.

Its impossible to know what the exact allocation of credit should be, so might as well split it 50/50. If you or anyone can point to studies that say otherwise, I'll gladly alter the formula.

Until then, 50/50 just seems like the epitome of fair lol.

Reply



Leave a Reply