tsherkin wrote:You're saying that AI included, the 08 Nuggets with a healthy Nene would look something like the average championship cast in the time given?
Indeed, the Nuggets are pretty close in terms of average championship cast in such a case and Iverson making a +3 impact on that team. That is the ballpark of Moses Malone or Charles Barkley. I think, people like to trash on Iverson, but overall he wasn't that much different than Moses Malone or Charles Barkley in overall team impact.
And Iverson was the best player on that team, not Anthony. Just keep in mind that Iverson had the highest +/- values on the Nuggets in 2008. When Anthony was out, the Nuggets looked as strong as with him, with the obvious exception of the b2b game on the road against the Hornets. Kleiza at SF worked well next to Iverson too. But when you have Camby, Nene, Martin and Najera in the frontcourt, it is most certainly not the average supporting cast anymore. Then we add Anthony and we are close to the ballpark of the 83 76ers or 93 Suns in terms of supporting casts. Which then tells us that Iverson is not a great pick to start a team, because he makes it difficult to find the needed fitting supporting cast players. 2001 would be another example of this for Iverson.
tsherkin wrote:The subjective elements of that roster arrangement don't line up, of course, because they were mismatched with their core star, but if that's the comment, then the level of talent certainly makes sense even if the skill distribution doesn't.
Do you want to imply that on championship teams the skill distribution was always perfect? There are always overlapping skillsets, where you have a position covered by two rather good players, while the backup for another position is rather weak. Thus, I don't see much of a problem with the Nuggets anno 2008 either. Again, the caveat is a healthy Nene here, who not just missed games, but also couldn't play up to his level.
tsherkin wrote:Well, I'm just curious, because most title teams of late haven't had two top 10 all-time players playing at an ultra-high level as the Lakers three-peat and the 12 Heat did, and the repeat titles that Kobe had seemed to have more overall talent than, say, the 03 Spurs or the 06 Heat or the 11 Mavs (though the Mavs were better than their lack of stars beyond Dirk indicates, of course). And really, you can look at the 05 Spurs as well.
A couple of points, the overall amount of stars is not necessarily a good indicator of the strength of the supporting cast. The Lakers 3peat was basically average without O'Neal despite having Bryant. The 2005 Spurs had a lot of talent with Ginobili playing on an insane high level, put that together with Tony Parker and we have a supporting cast actually exceeding the average playing level (they outscored their opponents while Duncan was off!). The 2006 Heat had a supporting cast on the level of the 2011 Mavericks, while playing worse with Wade than the 2011 Mavericks with Nowitzki. The 2010 Lakers were worse in terms of playing level than the 2008 Lakers, the 2003 Spurs were below average in terms of playing level with Duncan than an average championship team. So, there are a lot of things different for each of those championship casts and that is not even taking the 2004 Pistons or 2008 Celtics into account, two teams playing on a really high level while having a way better than average supporting cast. But when we look back, we see similar things in the past.
Overall, the important point is that when I select a player, I can't expect to get a 1983 76ers supporting cast as well or 1993 Suns or 2012 Thunder, it is just not likely. I have to look at the most likely supporting cast, and average level is what is most likely. How far can a player bring such an average casts is more important for me in terms of peak level play, than the ability to make a big impact on a weak cast or being able to sustain a +3 impact also on a +4 or +5 cast.
tsherkin wrote:I don't know, I think I'm just uneasy with the notion that you can look at an average supporting cast and create that as a neutral statistical environment from which to evaluate players because the nature of those casts is so quixotic and different between stars.
The last statement is true, and that is maybe the point where a lot of people seem to lose my train of thought. Because there are two major issues here: 1st, how easy can I get those necessary players in order to create such neutral environment for each player, and 2nd, how much difference does the respective player make in such a case. Both things are based on knowledge and assumptions, we know that a defensive big is easier to get than a offensive skilled big or a defensive wing easier than a defensive wing with great ball handling, we can get smaller ballhandler much more often than smaller players who can make plays from the mid to low post area, etc. pp. We also have informations about the impact of a player in a certain environment, while that environment most certainly is not statistical neutral.
tsherkin wrote:I suppose it is no different in concept than ElGee's whole SIO post from a little while ago on the stats board, though.
Indeed, it is pretty similar in the end. The only difference between Elgee and me seems to be, that I value impact on the average team more, while Elgee is focussed more on the high level championship teams, which at least from my perspective seems to be the cause of the differences.
tsherkin wrote:I don't know, maybe I'm picking at nits at this stage because I'm confused with what you're up to with your average cast, right?
Yeah, I think that's the case. The 1993 Suns have a bit of bad luck in terms of that, because they face a superior Bulls team, while in a different season they could have achieved more. In average a championship team is about +3.2 on off and +3.4 on def over the last 39 years (or +6.6 overall), the Suns had +5.3 on offense and +1.3 on defense, which again makes +6.6. The Suns 1993 had the average strength of a championship team, just not a good matchup in the finals with the Bulls being stronger than them.
tsherkin wrote:I'm seeing this as a seem between numerical analytics and contextual analytics
Without context the numbers are useless, that is an important point. If I have no clue how a player played and which skillset he had, I can't say much about that player. That is also the reason why I don't talk much about earlier players, because I have a hard time putting the numbers into context without having seen them play for an extended amount of minutes. We have some highlights or some game snippets with a focus on Chamberlain or Russell, but not for the most other players. We have a lot of hearsay and media stories, but as far as I'm concerned, those information are as biased as today. So, I'm very confident about the 2000's and 1990's players and have enough confidence about the 1980's and late 1970's, but everything earlier is usually something I keep my mouth shut.
The fact that I express my opinion with numbers might make it look as if that is entirely based on numerical analysis, but I think it is just way more precise to say a player is +3 or +5 than describing the player's skillset without giving a rather precise number (or even just a ballpark) how valuable such traits are supposed to be. Just saying "he is good at this" or "he is versatile" or "he is bad at this" doesn't actually mean a lot. Not all skills are equally important in the context of a 5on5 game, and not all boxscore numbers can be replaced in the same fashion. And when I talk about specific seasons and results, it is merely used as basis for information, not as the complete evaluation of the respective player. It would be naive to assume that a player would have the same result in a different environment.
As an example we can take the discussion about Garnett vs. Nowitzki, where the basic arguments are very similar (advanced boxscore metrics plus +/- based ratings) from the numerical perspective, the real hard facts are putting them both into the same ballpark, while there is an ongoing second discussion part which is based more on certain biases. There is a group of people arguing that Garnett's defense would be such a huge difference maker, that it is an easy decision, while another group takes the Mavericks higher success rate with Nowitzki than the Timberwolves with Garnett as the point of evidence as to why Nowitzki is clearly better suited for a top level team. Neither of that is useful, imho. At the end of the day, building a team around Garnett or Nowitzki for their average career seems to me similar easy, just that the Timberwolves completely failed at that, while at peak level Garnett was the better choice, even though the difference is not as big as some might want to believe. But without context I could easily fall into the same trap as the group of defense-supporters or team-success-supporters.
