Effects of League Expansion on League Strength?

[sansterre]

Standard Deviation of SRS is not a measure of league strength.

It's demographics of population - size of pool in playing age, international play, quotas in early years, playing coming into the league earlier, and impact of other leagues - not only ABA, but in early days AAU ball.

How would you define 'League Strength'?

[Jaivl]

Standard Deviation of SRS is not a measure of league strength.

It's demographics of population - size of pool in playing age, international play, quotas in early years, playing coming into the league earlier, and impact of other leagues - not only ABA, but in early days AAU ball.

How would you define 'League Strength'?

The definition is irrelevant, the point is that it's something probably not quantifiable by the usual team performance metrics. SRS stdev certainly seems to have nothing to do with it, either thinking logically or in practice.

[sansterre]

Standard Deviation of SRS is not a measure of league strength.

It's demographics of population - size of pool in playing age, international play, quotas in early years, playing coming into the league earlier, and impact of other leagues - not only ABA, but in early days AAU ball.

How would you define 'League Strength'?

The definition is irrelevant, the point is that it's something probably not quantifiable by the usual team performance metrics. SRS stdev certainly seems to have nothing to do with it, either thinking logically or in practice.

I'm sorry, but as soon as you say "the definition is irrelevant" you've lost me. If you are in a discussion about what determines 'League Strength' but you say "I don't care what that is", you pretty much have punted on the entire discussion.

I choose to define "League Strength" as "A balanced distribution of team quality, such that the worst teams aren't getting murdered and the best teams aren't crushing everyone".

This is obviously a completely arbitrary definition. But most of the others run into problems.

If you go with "League Strength = Great Teams" then that's fine, but Great Teams tend to be byproducts of weak under-classes of teams that feed their ratings. The '72 Bucks and Lakers were great . . . but much of what we think of as evidence of their greatness was actually their league being watered down like crazy. Take out the expansion teams and they show up as really, really good, but not stupid-great anymore. So the problem is that the "League Strength = Great Teams" weirdly often implicitly means "League Strength = lots of bad teams", which makes no sense.

So I choose to use standard deviations of SRS as a barometer for how artificially inflated the team stats for that year were. It's as good as I could think of as a way to try and compensate for that mechanic.

But seriously. If you can't define "League Strength" then discussing it *at all* is probably a waste of time.

[Jaivl]

How would you define 'League Strength'?

The definition is irrelevant, the point is that it's something probably not quantifiable by the usual team performance metrics. SRS stdev certainly seems to have nothing to do with it, either thinking logically or in practice.

I'm sorry, but as soon as you say "the definition is irrelevant" you've lost me. If you are in a discussion about what determines 'League Strength' but you say "I don't care what that is", you pretty much have punted on the entire discussion.

I choose to define "League Strength" as "A balanced distribution of team quality, such that the worst teams aren't getting murdered and the best teams aren't crushing everyone".

This is obviously a completely arbitrary definition. But most of the others run into problems.

If you go with "League Strength = Great Teams" then that's fine, but Great Teams tend to be byproducts of weak under-classes of teams that feed their ratings. The '72 Bucks and Lakers were great . . . but much of what we think of as evidence of their greatness was actually their league being watered down like crazy. Take out the expansion teams and they show up as really, really good, but not stupid-great anymore. So the problem is that the "League Strength = Great Teams" weirdly often implicitly means "League Strength = lots of bad teams", which makes no sense.

So I choose to use standard deviations of SRS as a barometer for how artificially inflated the team stats for that year were. It's as good as I could think of as a way to try and compensate for that mechanic.

But seriously. If you can't define "League Strength" then discussing it *at all* is probably a waste of time.

His point, not mine.

It's demographics of population - size of pool in playing age, international play, quotas in early years, playing coming into the league earlier, and impact of other leagues - not only ABA, but in early days AAU ball

Seems pretty clear he's referring to "league strength" as... well, the level of play of the league, in absolute terms. None of that is quantifiable by b-ref stats.

[DQuinn1575]

Standard Deviation of SRS is not a measure of league strength.

It's demographics of population - size of pool in playing age, international play, quotas in early years, playing coming into the league earlier, and impact of other leagues - not only ABA, but in early days AAU ball.

How would you define 'League Strength'?

I define League Strength as overall quality of players in the league.
The 1972 NBA was stronger than the 1972 ABA, although I'm guessing the Standard Deviation was higher in the NBA.
The quality of the players in the NBA in 1he late 70s can't have been that much different than early 80s - you can throw in drugs, and a few new names - but the 100th guy in the league in 1978 was probably somewhere around 105-ish maybe in 1981 - the late 70s didnt have super teams; I personally enjoyed it more when it was the Lakers, Celtics and 76ers in the early 80s rather than the Bullets, Sonics, Suns, Blazers in the late 70s - but thats a personal choice. I'd rather have 3 great teams in a league than 6 real good ones.

And the fact that the Sixers, Celtics, Lakers were so good, they probably caused at least part of that increase in the SD- if nothing changes except your top 3 teams get better, you SD grows - that was happening as the top teams in the league added Worthy, Moses, Walton, Ainge - they were adding more talent than the lousy teams.

Expansion dilutes the league - you add 3 teams with a combined minus 28 SRS to a 14 team league, and you've made the average 2 points worse - in rating players that is 0.3 points per 36 minutes - not a huge amount, but something. Remember though, the teams are protecting 7-8 players, and usually after 1 is selected they can protect 1 more. And they have traded or done other things to keep another player - so the playoff teams arent losing their top 7-8 players off the rotation, and arent impacted very much. What they get is a couple of extra wins, so maybe the 72 Lakers win "only" 66 games - they still were real darned good. The Bucks, Knicks, Bulls teams were still real good teams - the Knicks and Bucks were the last two champs, and each had great regular seasons. The Knicks had 2 top 100 players of all-time, and 3 good starters, the Bucks had 2 players who were considered by some at the time to be the 2 best ever, plus Dandridge. The Bulls had 4 real good players - I never ranked players too far down, but Love and Van Lier made 2nd team All-NBA at one point, Walker was 7 time all-star, Sloan was 2 time all-star, 6 time all-defense - maybe they all are Top 250?? of all time

[parapooper]

it's not rocket surgery:
Expansion
-> spread same player pool over more teams
-> guys start who would be bench players without expansion
-> bench gets filled up with guys who would be in acoounting or food service instaed of NBA without expansion
-> average/most teams get worse
-> stars have inflated stats

[penbeast0]

it's not rocket surgery:
Expansion
-> spread same player pool over more teams
-> guys start who would be bench players without expansion
-> bench gets filled up with guys who would be in acoounting or food service instaed of NBA without expansion
-> average/most teams get worse
-> stars have inflated stats

But HOW MUCH inflation; now you are getting into the interesting areas . . . .

[euroleague]

Thank you for post this ty . . . I hope Odinn hasn't turned this back into a Jordan thread because it's an interesting question and one we have a number of opinions on.
).

Most of the data is pretty obviously wrong though.

Look at the data from the 2000s - it says like 3 teams were below 40%? Philadelphia was below 40% 3 times in the 2010s.

Taking franchise averages is a pretty absurd way to calculate league strength. There could be 10 teams dominating, with lots of frothiness as to which those teams are, and the rest of the league can suck... but, as long as those 10 teams change between different 3 franchises - like LBJ - this statistic will not show that team below 40%.

In conclusion, the way this data is structured is designed to support an agenda, and isn't really helpful

In the modern era, we have tanking like never before. And free agent movement like never before. Because there wasn't FA movement in the early eras, certain teams were stuck at lower win percentages more.

I just don't know how his data can be right though

[magicman1978]

for inflated stats - wouldn't that be highly reflected in something like PER since it's normalized to league average? And if we believe some stats were inflated due to expansion, how do we reconcile that with the "inflated" numbers of today's stars as there's obviously more talent today?

[ty 4191]

Most of the data is pretty obviously wrong though.

Look at the data from the 2000s - it says like 3 teams were below 40%? Philadelphia was below 40% 3 times in the 2010s.

Taking franchise averages is a pretty absurd way to calculate league strength. There could be 10 teams dominating, with lots of frothiness as to which those teams are, and the rest of the league can suck... but, as long as those 10 teams change between different 3 franchises - like LBJ - this statistic will not show that team below 40%.

In conclusion, the way this data is structured is designed to support an agenda, and isn't really helpful

In the modern era, we have tanking like never before. And free agent movement like never before. Because there wasn't FA movement in the early eras, certain teams were stuck at lower win percentages more.

I just don't know how his data can be right though

Hello Euro,
I have no "agenda" whatsoever.

I simply wanted to ask the scribes and experts here about league strength over time. And its implications on player and team records, particularly.

As to the data? It is overall franchise records during given spans of time. I used stathead (basketball reference's subscription play index) to find overall franchise records in given timeframes.

Example: Only 3 out of 30 teams have had an OVERALL winning percentage of .400 or below (out of 30 franchises) in the past 10 seasons:

-Phoenix (306-498) (.386 WPCT)
-Minneapolis (292-494) (.372)
-Sacramento (293-501) (.369)

Now contrast that with 1989-1998. An, yes, it happens to nearly be a certain Basketball God's entire prime and all his championships.

29 total teams:
-NJ .398
-WAS .388
-DEN .387
-SAC .367
-LAC .359
-DAL .323
-MIN .321
-TOR .272
-MEM .195

Contrast that with 1979-1988. Pre major expansion, which killed league parity and added a ton of bad players to the league:
23 total teams:
-GSW .399
-IND .399
-CLE .365
-LAC .339

[magicman1978]

All of us have some sort of agenda/bias, nothing wrong with that (just need to keep an open mind in discussions)- it's clear in your last post what yours is. Assessing overall league strength is different than assessing how it impacts certain players (Talent is not evenly distributed and players don't play the same opponents when it truly matters most). To truly have a relevant discussion, you're going to have to divorce the two.

[penbeast0]

How would you define 'League Strength'?

I would define it as the quality of the average player minute in the league. It's not the overall quality of the league defined as a sum; a bigger league made bigger by adding worse player to the bottom of it to me lessens the strength of the league. So, it would be some measure of average quality of a player in the league.

I would go a bit further and say that if player A plays 3000 minutes in the league and player B gets a tenday with less than 100 minutes, player A's quality should help define the strength of the league more. Thus I break it down to average player minute which means that the superstars (healthy ones) do determine more about the quality of the league than the scrubs per player because they play more minutes.

[feyki]

We can't measure the talent and bell curve system is not open to calculate any.

[ty 4191]

Post redacted -- discussing poster not content.

[parapooper]

it's not rocket surgery:
Expansion
-> spread same player pool over more teams
-> guys start who would be bench players without expansion
-> bench gets filled up with guys who would be in acoounting or food service instaed of NBA without expansion
-> average/most teams get worse
-> stars have inflated stats

But HOW MUCH inflation; now you are getting into the interesting areas . . . .

it depends on how steep the player-quality curve from #1 to #15 (and beyond the league) was before the expansion. I don't see how there could be no negative effect on average team quality, but with so many other things going on it would be insanely hard to quantify. There would be a lot of background noise as well, for instance a team could easily get better after dilution by for instance being forced to replace a starter the coach liked with a better bench player (sth like Draymond starting to get minutes).

[parapooper]

for inflated stats - wouldn't that be highly reflected in something like PER since it's normalized to league average? And if we believe some stats were inflated due to expansion, how do we reconcile that with the "inflated" numbers of today's stars as there's obviously more talent today?

Not sure stats are inflated today, but if they are it's obviously other factors at work. I'm pretty sure if today's stars played more minutes against bench players and G-league guys their stats averaged over the season would look better, and today's talent pool is pretty deep compared to NBA-expansion times.
But yes, stat inflation would be vs what they would be without expansion, not necessarily vs. other times in NBA history where other factors also change. For instance a trend to more star-centered play/higher star usage would also inflate star stats - their stats would be scaled against peers who look statistically worse because they don't have the ball while in the expansion scenario peers would actually be worse. Statistically the effect would be similar.

[magicman1978]

for inflated stats - wouldn't that be highly reflected in something like PER since it's normalized to league average? And if we believe some stats were inflated due to expansion, how do we reconcile that with the "inflated" numbers of today's stars as there's obviously more talent today?

Not sure stats are inflated today, but if they are it's obviously other factors at work. I'm pretty sure if today's stars played more minutes against bench players and G-league guys their stats averaged over the season would look better, and today's talent pool is pretty deep compared to NBA-expansion times.
But yes, stat inflation would be vs what they would be without expansion, not necessarily vs. other times in NBA history where other factors also change. For instance a trend to more star-centered play/higher star usage would also inflate star stats - their stats would be scaled against peers who look statistically worse because they don't have the ball while in the expansion scenario peers would actually be worse. Statistically the effect would be similar.

Lets test the theory. Can we agree that PER is a good baseline since it's normalized to league average? I'm going to pick a random PER of 22 as the bar (rounding up). If I understand it correctly. PER is like grading on a curve. Add more bad students and the good students should stand out more. But if you add more good students, then the previous good students would not stand out as much.

2021 - 31 players (13 at 25+)
2020 - 25 players (11 at 25+)
2010 - 15 players (5 at 25+)
2000 - 12 players (6 at 25+)
1997 - 14 players (4 at 25+)
1996 - 14 players (9 at 25+) - expansion
1995 - 11 players (5 at 25+)
1994 - 12 players (3 at 25+)
1992 - 13 players (4 at 25+)
1991 - 13 players (5 at 25+)
1990 - 15 players (7 at 25+) - expansion
1989 - 11 players (4 at 25+) - expansion
1988 - 11 players (3 at 25+)
1987 - 10 players (4 at 25+)
1977 - 5 players (1 at 25+)
1976 - 5 players (2 at 25+) - expansion
1975 - 5 players (2 at 25+) - expansion
1974 - 4 players (1 at 25+)
1973 - 5 players (2 at 25+)

Based on that quick/dirty analysis, nothing jumps out to me about star stats being inflated due to expansion - maybe a small bump in an expansion year, but it goes down the very next year (Abd there's no way talent levels are normalized in just a year). The deviation is pretty consistent. There was a jump from the 70s to the 80s and was relatively consistent up until the last 5 years or so.

[parapooper]

for inflated stats - wouldn't that be highly reflected in something like PER since it's normalized to league average? And if we believe some stats were inflated due to expansion, how do we reconcile that with the "inflated" numbers of today's stars as there's obviously more talent today?

Not sure stats are inflated today, but if they are it's obviously other factors at work. I'm pretty sure if today's stars played more minutes against bench players and G-league guys their stats averaged over the season would look better, and today's talent pool is pretty deep compared to NBA-expansion times.
But yes, stat inflation would be vs what they would be without expansion, not necessarily vs. other times in NBA history where other factors also change. For instance a trend to more star-centered play/higher star usage would also inflate star stats - their stats would be scaled against peers who look statistically worse because they don't have the ball while in the expansion scenario peers would actually be worse. Statistically the effect would be similar.

Lets test the theory. Can we agree that PER is a good baseline since it's normalized to league average? I'm going to pick a random PER of 22 as the bar (rounding up). If I understand it correctly. PER is like grading on a curve. Add more bad students and the good students should stand out more. But if you add more good students, then the previous good students would not stand out as much.

2021 - 31 players (13 at 25+)
2020 - 25 players (11 at 25+)
2010 - 15 players (5 at 25+)
2000 - 12 players (6 at 25+)
1997 - 14 players (4 at 25+)
1996 - 14 players (9 at 25+) - expansion
1995 - 11 players (5 at 25+)
1994 - 12 players (3 at 25+)
1992 - 13 players (4 at 25+)
1991 - 13 players (5 at 25+)
1990 - 15 players (7 at 25+) - expansion
1989 - 11 players (4 at 25+) - expansion
1988 - 11 players (3 at 25+)
1987 - 10 players (4 at 25+)
1977 - 5 players (1 at 25+)
1976 - 5 players (2 at 25+) - expansion
1975 - 5 players (2 at 25+) - expansion
1974 - 4 players (1 at 25+)
1973 - 5 players (2 at 25+)

Based on that quick/dirty analysis, nothing jumps out to me about star stats being inflated due to expansion - maybe a small bump in an expansion year, but it goes down the very next year (Abd there's no way talent levels are normalized in just a year). The deviation is pretty consistent. There was a jump from the 70s to the 80s and was relatively consistent up until the last 5 years or so.

Thank you, that's interesting and unexpected
Seems weird intuitively but it looks like no significant stat inflation due to expansion. Maybe stars were playing down to competition? Or worse players were making up for it by being less lazy on defense? Star players just happening to be lower quality on 3 separate expansion occasions is a very unlikely explanation.

Reagrding today's stat inflation (or inflation of more guys with great stats) it's probably mostly in per minute stats because stars play less and can play harder over short stretches and have higher usage/dominance because of less in-game rest needed.
But who knows, guess things are just complicated

[euroleague]

Most of the data is pretty obviously wrong though.

Look at the data from the 2000s - it says like 3 teams were below 40%? Philadelphia was below 40% 3 times in the 2010s.

Taking franchise averages is a pretty absurd way to calculate league strength. There could be 10 teams dominating, with lots of frothiness as to which those teams are, and the rest of the league can suck... but, as long as those 10 teams change between different 3 franchises - like LBJ - this statistic will not show that team below 40%.

In conclusion, the way this data is structured is designed to support an agenda, and isn't really helpful

In the modern era, we have tanking like never before. And free agent movement like never before. Because there wasn't FA movement in the early eras, certain teams were stuck at lower win percentages more.

I just don't know how his data can be right though

Hello Euro,
I have no "agenda" whatsoever.

I simply wanted to ask the scribes and experts here about league strength over time. And its implications on player and team records, particularly.

As to the data? It is overall franchise records during given spans of time. I used stathead (basketball reference's subscription play index) to find overall franchise records in given timeframes.

Example: Only 3 out of 30 teams have had an OVERALL winning percentage of .400 or below (out of 30 franchises) in the past 10 seasons:

-Phoenix (306-498) (.386 WPCT)
-Minneapolis (292-494) (.372)
-Sacramento (293-501) (.369)

Now contrast that with 1989-1998. An, yes, it happens to nearly be a certain Basketball God's entire prime and all his championships.

29 total teams:
-NJ .398
-WAS .388
-DEN .387
-SAC .367
-LAC .359
-DAL .323
-MIN .321
-TOR .272
-MEM .195

Contrast that with 1979-1988. Pre major expansion, which killed league parity and added a ton of bad players to the league:
23 total teams:
-GSW .399
-IND .399
-CLE .365
-LAC .339

Taking an average over a 10 year span is pointless, because those are completely different teams. The only thing that's the same is the franchise....

The way you're using statistics makes no sense.

[ty 4191]

Taking an average over a 10 year span is pointless, because those are completely different teams. The only thing that's the same is the franchise....

The way you're using statistics makes no sense.

The thread is about LEAGUE strength. So yes, franchise records are what matter over a staisitically significant sample of years.

1979-1988 is before 6 bad or terrible expansion teams were added to the league, 1989-1998 was during and right afterwards.

The percentage of teams below .400, relatively speaking, is illustrative of how much more balanced the 80's were than the 90's.

And, do you think it's a coincidence that the NBA hasn't expanded once since 2004, and that only 3 out of 30 teams are below .400. overall, during that entire timeframe? And, also, during the last 10 years?

Expansion destroys league quality and it destroys competitive balance.