It's spelled with an r, not an n

[schtevie]

Here's another try, a final one.

Each CBA is structured the way it is. In implementation, players negotiate salaries, and the determination of these are determined by many factors: the text of the CBA (that imposes certain price ceilings and floors, amongst other things), the willingness of owners to pay, the particular, local market conditions faced by particular players at particular times, AND also "skill". (And I've probably missed some others.)

What I am saying - and I am a bit surprised that it is controversial - is that there is a positive relationship between salaries and skill and that it can be divined from the data, and that is because there are some market forces determining salaries. I took Daniel's formula as a first cut of this, and specifically stated that it cannot be considered the last. Then the inference, based upon the large, Maverick overpayment that this formula suggested, was that it is therefore reasonable to believe that the revealed return to analytics, if positive, could not be very large.

What I did not say is that the CBA and the structure of the league created an approximate free market for labor. This is the straw man of mystic's imagination. Nothing against creative and active imaginations, but it is simply not true to claim that actual salary data, in the context of actual legal and competitive conditions, cannot be used to infer the price of "skill", all else equal.

[Bobbofitos]

Here's another try, a final one.

Each CBA is structured the way it is. In implementation, players negotiate salaries, and the determination of these are determined by many factors: the text of the CBA (that imposes certain price ceilings and floors, amongst other things), the willingness of owners to pay, the particular, local market conditions faced by particular players at particular times, AND also "skill". (And I've probably missed some others.)

What I am saying - and I am a bit surprised that it is controversial - is that there is a positive relationship between salaries and skill and that it can be divined from the data, and that is because there are some market forces determining salaries. I took Daniel's formula as a first cut of this, and specifically stated that it cannot be considered the last. Then the inference, based upon the large, Maverick overpayment that this formula suggested, was that it is therefore reasonable to believe that the revealed return to analytics, if positive, could not be very large.

What I did not say is that the CBA and the structure of the league created an approximate free market for labor. This is the straw man of mystic's imagination. Nothing against creative and active imaginations, but it is simply not true to claim that actual salary data, in the context of actual legal and competitive conditions, cannot be used to infer the price of "skill", all else equal.

The point mystic is making is that Dallas may have overpaid for wins, based on your definition, but that doesn't mean their analytics were worth less. Because they were willing to overpay, it allowed them to get a good player (terry) IF they took along a bad player (henderson). The aggregate, in a pure dollar/win relationship, looks like a loss.

[mystic]

The point mystic is making is that Dallas may have overpaid for wins, based on your definition, but that doesn't mean their analytics were worth less.

Indeed, that's my point.

Because they were willing to overpay, it allowed them to get a good player (terry) IF they took along a bad player (henderson). The aggregate, in a pure dollar/win relationship, looks like a loss.

There were willing and actually forced by the CBA to pay more, otherwise they couldn't have improved the team. And they constantly improved the team until a certain point when they had to spend money in order to keep the level due to players declining (aging) or getting injured.
In fact the NBA is setup in order to give the weaker team a better chance to improve while making it more expensive for better teams to stay put or actually improve. And that is not a linear relationship either, if we take the luxury tax into account. The current CBA is actually so harsh for such teams, that the Mavericks changed their approach in order to have a chance to constantly be among the top teams.

[mystic]

What I did not say is that the CBA and the structure of the league created an approximate free market for labor. This is the straw man of mystic's imagination.

When did I say in any of my posts, that you said the league would be a free market? I NEVER wrote anything close to that. I just simple pointed out that the assumption you made is only working in a free market and not in a cartel like the NBA. Especially, and that is the MOST important point, when the CBA is written in the way that it should be more expensive for better teams to improve. That is a big issue, an issue you seem to ignore.

Nothing against creative and active imaginations, but it is simply not true to claim that actual salary data, in the context of actual legal and competitive conditions, cannot be used to infer the price of "skill", all else equal.

Sorry, but the price is not just determined by skill. The price is determined by demand and supply, and some player types are just simply more expensive, because they are not often available. But in order to achieve to be a constant good team, you need such players. Thus, you have to pay a higher price just to get those specific players. Dampier wasn't paid by the Mavericks, because he was thought to be such high impact player, he was signed, because a team needs the service of a center like Dampier. What you are implying is that they could have just signed a cheaper +1 player, while ignoring that this cheaper player would likely be a guard or wing player, while the Mavericks had already covered that position. And when your expensive centers are getting injured (Bradley, LaFrentz, Dampier), you are in need of signing another one by "overpaying" again. That's how Diop ended up getting a MLE contract from the Mavericks, because Dampier showed that he is not constantly available. But what kind of other options did the Mavericks had in order to get a backup center? That's what you need to show here, that the Mavericks had the opportunity to sign different players, but choose to sign a specific player because the analytic department said so. Can you give any specific examples here? Can you prove that a different set of players would have achieved the same success while being cheaper and available for the Mavericks? That's what you would need to show in order to assume how less value analytics offer to the teams.

[Crow]

I think they're a team very much on the rise, and that's large part because of BC.

Bargnani and DeRozan have always been estimated negative on RAPM. Will that change with the new roster additions? Last chance to turn the corner with them, IMO. Maybe new smaller roles will help. If they try it.

[Bobbofitos]

I think they're a team very much on the rise, and that's large part because of BC.

Bargnani and DeRozan have always been estimated negative on RAPM. Will that change with the new roster additions? Last chance to turn the corner with them, IMO. Maybe new smaller roles will help. If they try it.

Bargs was a terrible contract decision, bad #1 pick, and honestly just a total head scratcher. But most GMs have that #1 love affair, and he's obviously BC's love. Maybe Bargs will turn the corner w/ a full season of health and Dwayne Casey in charge. We'll see.

As far as DeRozan: I think he's performed above value relative to his rookie contract. Yes, a negative player, but he's been a legit NBA player who has been forced to fill a lot of minutes. I don't think he was a bad pick at all. The real question is whether or not he gets an extension... That's the bigger test/challenge for BC.

[Crow]

Worst year for DeRozan on almost every major boxscore stat and winshares per 48. But it appears his defense improved from horrible. His non-prior informed RAPM was nearly neutral. One might be semi-optimistic I guess. His usage rate is absurd given his efficiency though.

[Bobbofitos]

Worst year for DeRozan on almost every major boxscore stat and winshares per 48. But it appears his defense improved from horrible. His non-prior informed RAPM was nearly neutral. One might be semi-optimistic I guess. His usage rate is absurd given his efficiency though.

He's been paid 7.3 mil thus far in his career. Winshares credits him for 8.1 ws. On a dollar to wins basis, he's performed great. No?

[Crow]

In briefly referring to him, I was focused on simply court impact. I appreciate your argument about "value" given his rookie scale contract and very high minutes. He is a current value that I would not sign to a big contract though or probably even a "fair value" contract that rewards for minutes received. I think there will be better options and better values. He had the 19th worst offensive rating (off rtg of 100) of the 146 guys to play 25+ minutes per game in 40+ games last season. http://bkref.com/tiny/eCmp2

[Bobbofitos]

In briefly referring to him, I was focused on simply court impact. I appreciate your argument about "value" given his rookie scale contract and very high minutes. He is a current value that I would not sign to a big contract though or probably even a "fair value" contract that rewards for minutes received. I think there will be better options and better values. He had the 19th worst offensive rating (off rtg of 100) of the 146 guys to play 25+ minutes per game in 40+ games last season. http://bkref.com/tiny/eCmp2

I think underlined is key. Basically, relative to his draft slot and the on-court minutes played, although DeMar has hardly been "good", I think he's outperformed his draft position as far as total production. The real decision will be when an extension is in the works. If the Toronto FO "reward" DeMar for gaudy per game numbers (they aren't even THAT gaudy!) then that speaks to a larger, general office/analytic mistake. So, it's sorta a question of how highly they view his current/recent past production.

[Crow]

Ainge called Rondo his best player.

Prior-informed RAPM last season had Garnett with 5.6 times the positive impact. 10 yr RAPM has Garnett at +8 and Rondo at slightly below neutral. Good thing Garnett re-signed. Boston lead by Rondo without Garnett is likely not going to be much in a few years. Might they try to pump up Rondo and dump him? I guess I would.

Played great in the playoffs? He had the best ws/48 on the team (not that hard, only took a .155) but his shooting and scoring efficiencies were nothing special.

[DSMok1]

Be wary of prior-informed RAPM when there are potentially significant aging effects in play, since there is no aging adjustment applied between seasons.

That said--RAPM is still making a very strong statement in this case, and far beyond any potential issue caused by aging.

[Crow]

True. I am generally looking at both prior-informed and non-prior informed these days. This time I added 10 year but forgot to write about the non-prior informed. Garnett only 2.4 times better on that last season.

Rondo peaked in 2009 and has never been above +2 on prior-informed. Nice but not a super-star. Maybe not even a star, depending on where you draw the line. Over 50 players were above +2 on prior-informed last season. Are all of them stars?

http://stats-for-the-nba.appspot.com/pm/553.html
Just 1 point better than Avery Bradley when playing with the same 4 others? That is what their overall RAPM estimates say too.

Only 1 of the top 4 Boston playoff lineups with Rondo at the controls broke 100 on offensive efficiency. In the regular season it was two above league average, two below.

[Mike G]

Ainge called Rondo his best player.

Prior-informed RAPM last season had Garnett with 5.6 times the positive impact. ..

How does RAPM give such a ratio?
Are you converting RAPM to "positive impact" somehow?

[DSMok1]

Ainge called Rondo his best player.

Prior-informed RAPM last season had Garnett with 5.6 times the positive impact. ..

How does RAPM give such a ratio?
Are you converting RAPM to "positive impact" somehow?

Right, you can't say it's a multiple. Crow should be looking at difference, not ratio.