APBRmetrics

[Dennis_D]

Here is a system I have developed from evaluating the drafting of 17 GM's. Please give me some feedback.

Here's my GM rating (best to worst):
+4.1 - Rod Thorn of New Jersey
+4.0 - Geoff Petrie of Sacramento
+3.9 - Randy Pfund of Miami
+3.8 - John Paxson of Chicago
+3.1 - Bryan Colangelo of Phoenix
+3.0 - Don and Donnie Nelson of Dallas
+2.6 - Mitch Kupchak of the LA Lakers
+2.1 - Joe Dumars of Detroit
+1.6 - Indiana
+1.1 - Rick Sund of Seattle
-0.4 - Kevin McHale of Minnesota
-0.7 - RC Buford of San Antonio
-1.6 - Jerry West of Memphis
-2.1 - Carroll Dawson of Houston
-5.2 - Kevin O'Connor of Utah
-8.3 - Elgin Baylor of LA Clippers

I expect for each pick in the draft order to net a certain level of talent. If a GM drafts better talent with lottery picks than another GM does with late first round picks, that GM isn't necessarily doing a better job. So, I defined what level of talent I expect with each pick, then I will give GM credit (positive points) or blame (negative points) if he the player he drafts turns out better or worse than I expected. Here is the scoring chart:

[Mountain]

The sliding credit conceptual framework makes sense but here is one opinion about some particulars:

Getting an all-star is still an achievement even very early so I might make that at least a +2 at the start of the table, maybe even +3. (And shift at least some of the others accordingly.)

There will always be cases of players who fall between categories but I'd suggest below average starters get credit halfway between a solid starter and a rotation guy. I think that is the spot where splitting hairs further would be most useful.

After rotation moving to the right, I might chop the size of the credit steps to 1/2 point. To me it matters less after rotation and is often a lot about limited opportunity and less about major gradations in talent level.

I might look at reducing the change for moving down the table a row to .33 or even .25, or perhaps start at .5 and slowly shifts to lower or perhaps reduce the number of rows. I feel picking late gems or even just better than expected is maybe getting too much credit compared to just picking a gem period or even solid value for the slot, which is still hard. Two all-stars from 5-8 equal one star at 30? I'd take the former. It is both a far better circumstance but, apart from that, I think it is also a more powerful, better performance. A bench player at 41+ equal to a star at 13-16? I'd take the later. I'd consider disrupting the consistency of the steps further beyond the above suggestions if subjectively I felt it was warranted to get different picks of equal credit closer to feeling equal across the many spots on the table.

Welcome, thanks for sharing the ratings and hope the comments that come help.

[Rasta978]

Dennis,

Very interesting analysis. You've obviously put a lot of time and thought into this. Great contribution to the board.

Any chance you could score John Gabriel, gm of the Magic from 1993 through 2003? I always considered him a terrible gm who wasted draft picks on the likes of Gertt Hammink, Brooks Thompson and Rodney Dent. I'd be curious to see how you rate him.

[THWilson]

I enjoyed this, like the concept.

I would re-jigger some of the player rankings. Barbosa is a high contributing sixth man on a very good team, Gomes, Collison and Marvin Williams start on very bad teams. If the rankings were more rooted in contribution, rather than minutes, you could weight these players more appropriately.

Win Shares seem like a perfect measure since it combines minutes with contribution.

[Ben]

[Jakedfw]

I think that you really need to add a scoring system for non-drafted free agents. For example, Dallas signed Marquis Daniels as a non-drafted free agent, and he was second team all-rookie his rookie year. Miami signed Udonis Haslem as an undrafted free agent, and he went on to start on their NBA title team.

[LanierFan]

The scoring classification seems a little mushy to me. All Stars can be easier or harder to draft depending on the conference you're in, and the "Star" category punishes GMs for doing the rest of their job well -- i.e., creating a lineup of above-average veterans via trade or free agency.

One other point: excluding Mehmet Okur is a head-scratcher for me. You say that Detroit "didn't recognize" him to be a steal, which is ridiculous. What they recognized was that, prior to the closing of the Arenas loophole, they couldn't keep Memo and the player who'd brought them the 2004 championship, Rasheed Wallace. That was an unusual situation, but one can control for it in far more reasonable ways than arbitrarily deducting 60% of his value.

[Dennis_D]

Corrected the ratings of Okur and Dahntay Jones

[Ben]

I decided to try out my proposed method of comparing the nth draft pick to the nth best player in the draft as rated by career win shares. Note that this has the odd effect of a GM never getting positive credit when they have the number one pick.

To illustrate with an example, take Joe Dumars with the number 2 pick in the NBA draft, we compare his pick of Darko (15 win shares) to Dwyane Wade (111 win shares) to get a score of -96.

When a player was traded before he ever played, I decided to grade the GM of the franchise where the player actually ended up. Thus, Gabriel is graded on Penny rather than on Webber. In this particular instance, the value is zero for both cases, so it didn't have any effect on Gabriel's rating.