APBRmetrics

[Chicago76]

Okay then. If you want to look at the possibilities empirically rather than logically, then don't exclude one of the 4 based upon logic.

option 1/2: during contraction, talent is either diluted or concentrated
option 3/4: during expansion, talent is either diluted or concentrated

You have cited one contentious claim: (2) - talent dilution but league quality concentration or non-change - is indeed a logical possibility.

However, so is an alternative outome: talent contraction, with league dilution. Look back at 76/77. What occurred? Contraction. What also occurred simultaneously? The ABA/NBA generally had lost several key players with no one around to prop up the league. There was a young crop of Marques Johnson, Parish, Dantley, etc, however when you compare them to what was lost in the last 3-4 years, they pale in comparison at the age of 22. Net effect, I would argue the talent was concentrated in 76/77, however, this was only the case because the degree of contraction exceeded the degree of talent loss. So in some "historical sense" we have already seen the possibility you have excluded based upon logic. You can't exclude one scenario logically but include another that fails a similar logic test if both cut the mustard empirically.

You state the following:

I define talent as I believe it is conventionally defined, as the quality of the "average" NBA player unweighted by playing time. Is this not what folks typically mean when they decry the deleterious effects of expansion, players of lesser POTENTIAL being necessarily allowed into the league?

You can't simply unweigh playing time because otherwise, a star player would never exit a game. In other words, a star playing 4 ten minute rotations is generally worth less in the last minute, when he is tired, than he is in earlier in the rotation. If this were not the case, that star player would never leave the court unless he was in foul trouble.

If you want to empirically look at expansion/dilution, look at all players and compare their absolute output (efficiency times minutes) pre-expansion to their absolute output post-expansion.

By running this empirical test, it is relatively confirmable that expansion leads to dilution, and contraction leads to concentration. Talent leaps happen. But in a relatively mature sport, you can not empirically witness a 4-5% improvement in quality that more than offsets a 3-4% talent dilution due to league expansion. One force may work against the other, however, league expansion ultimately wins out. If this were not the case, then you could empirically demontrate the evidence against this general observation.

Empirically, when expansion occurs, the players playing in the pre-expansion league contribute more in the post expansion league. There are other factors at play, but relatively speaking this would be like a sailor using wind speed and his fart in the wind to predict boat velocity: One is important, while the other is generally much less important.

This could change, because as the league continues to grow, the dilution effect could be less substantial, however, again, empirically, we have not seen this to be the case.

[Mike G]

[Mike G]

(part II)
Rather than just look at minutes, I'm looking at a specific strength: Rebounding. Someone's going to get a rebound, they're easier to get when the opponent is weaker, and the average player doesn't lose a lot of rebounding (%) from age 24 to 28.

A 'rebound rate' of sorts was generated for each player, based on average Reb/G by NBA teams that year. This is crude but probably accurate when averaged over a large sample. To avoid confusion with conventional RebRt, I'm dropping the decimal (so it's Reb per 10,000 available). Average is 1000 for all seasons.

Expansion seasons are noted after the regular columns. The period from 1955-61 held just 8 teams.

[schtevie]

Chicago76, I am pretty sure that I got my logical categories correct, and I defined these in the context of a league expanding. Similar categories would obtain for league contraction.

Anyway, I am not really sure I understand your point about 1976/77. What had been 27 teams within two leagues, shrunk to 22 teams in one league. The expectation then is that marginal talent would have been expelled from professional basketball. So, talent (by my unweighted definition) should have concentrated. And, if I understand you correctly, you say that, in addition, new entrants to the league were of above-average promise (as realized by their careers), so then this too would argue for talent concentration. So, what's the problem? League contraction coincided with talent and league quality concentration.

Regarding your dissatisfaction with the idea of separating the definition of talent (as an unweighted quality measure) from league quality (as a playing-time weighted measure) I am not saying that expansion isn't prone to league quality dilution; I am saying that teams can substantively mitigate against this.

Take the expansion of 1987/88 to 1988/89. The NBA increased the number of teams from 23 to 25, if I recall correctly. This was an 8.7% increase in expected player minutes. Now, I have not looked at possible mitigating effects of changes in the age-distribution of the league, nor the marginal effect of the quality of the 1988/89 rookie class, but I take your point that, all else equal, the league should in a sense have decreased 8.7% in quality.

However, the fact of the matter is that most of the impact players (think of these as the top 20% to 30% of players with positive adjusted +/-) could easily increase their playing time by 8.7%, offsetting the talent diluting effects of the expansion. So a 35 minute player becomes a 38 minute player. Big deal.

In fact, I think that this is what you are picking up as happening with your WS data, is it not?

And a brief reply to Mike G's rebounding argument. I don't get it. I would not expect that rebounding would be the relevant statistic to manipulate in search of finding talent or league quality dilution. Rebounding is rebounding. If a shot is missed, it will be rebounded, no matter what the quality of players. What am I missing?

Again, as I understand the conventional argument, it is offensive skill (scoring, perhaps passing) that is relatively scarce and whose average suffers with league expansion.

As a final point, I think it is a lot more helpful to think of expansion/dilution issues, not in terms of game statistics, but more broadly in terms of population variables.

If the relevant domestic population feeding the NBA (breaking it down - or not - by racially or socio-economically sub-groups) grows at X% per year, this means (all else equal) in expectation that the talent level of each rookie class (taking into account the changing rules for minimum age requirements for league entry) is increasing by that same X%. This is the demographic reality. That there aren't more feeder institutions to process the greater population is not the point. They get choosier too.

Add then in the more difficult to measure but conceptually similar factor of foreign entrants, and the benchmark is that the league significantly concentrates in talent over time, interrupted only by the effects of expansion.

Confusing what should be this clear picture are changes in the way the game is played over time. But this fact is likely independent from league expansion (or if not, a separate argument is required to link them).

[Mike G]

[gabefarkas]

[Mike G]

Is it being suggested that population growth, in and of itself, drives talent growth in a specific area like basketball? Does anyone assume the talent in soccer, hockey, lacrosse, chess, horseshoes, or horse-shoeing, are increasing at the same rate?

Doesn't every activity require some actual activity, to gain proficiency? Are all (or any) of the above-mentioned activities gaining on such pastimes as video gaming? After school and video games, how many hours does an adolescent have left for sports?
_________________
40% of all statistics are wrong.

[Mike G]

Extended and expanded. Having dug up the relevant figures for 2004 thru '07, here are alternative 'cumulative league strength indices'.

Because the aggregate (1952 thru 2007) rebound rate carryover (ages 24-27 in previous year to ages 25-28 in reference year) is not exactly 1, in the 'first pass', I guessed a correction factor of .995 as the 'expected' rate. I may or may not be able to explain this.

In any case, the first pass revealed peaks and valleys in the strength of the NBA. As with being on a real mountain peak, you can't really tell whether a distant peak is higher or lower than the one you're on. You can, however, tell that the ground in between is lower.

The local topography is sometimes of more import than the long-distance, absolute difference. When I change the RebRt expectency by tiny increments, large changes are seen in the accumulation. You may take your pick of the columns that best represent your view of the evolving NBA.