APBRmetrics

I admit, this is a nerdy post. I tried to condense it for discussion purposes--I could write an entire book on my findings. Anyhow, I thought I'd share my thoughts...


The prevailing theory among NBA purists is "league expansions" reduce the quality of competition. Bill Russell argues this point in multiple interviews romanticizing his playing career. Russell claims, as do others, that minimizing league franchises restricts roster spots to only high quality players. This claim overlooks other moving parts such as the effects of commissioned rule changes on quality of play, but moreover Russell and others are simply comparing competition quality in a vacuum. They're comparing the competition quality in the late 90s (thirty team league) to the early 60s (nine team league) without considering the other key variables. It's seems somewhat paradoxical to explain how a league--gradually overcoming prejudicial barriers, optimizing recruitment and training practices, expanding into new markets and increasing average salaries over 900% between 1968 and 1998, couldn't possibly produce a lesser quality product over those thirty years if it added 40 additional teams.

My work seems to reflect that concept, in 1966 the NBA featured nine teams, yet in 1969, after inducting five new franchises and with incoming competition from the newly formed ABA in 1967, the league quality 12% by the season ending 1969. Competition quality increased the following season by 3.3%, even with the departure of Celtics great Bill Russell. The 1969-70 season was the debut of arguably the greatest Center of All-time--Kareem Abdul-Jabbar.

In 1966 four Centers met my threshold for high quality play (my criteria is based on an all-time metrics called value shares): Chamberlain, Russell, Bellamy, and Beaty. In 1969 there were seven: Chamberlain, Reed, Russell, Beaty, Unseld, Rule, and Hayes. The NBA may have added five franchises in three seasons beginning in 1967, but the quality performance of the League's big men more than outweighed the increasing roster availability effect on competition quality.

It's easy to claim that expansion created an easier path to productivity for Chamberlain, Russell, Bellamy, and Beaty compared to season ending 1966. I just don't see where that's the case. The average Center productivity in season ending 1969 increased 20% overall compared to the 1966 season. Coach Holzman took over the Knicks job midway through the 1968 season, his utilization of guards Frazier and Van Arsdale to apply pressure defense helped to relive Bellamy and Reed in the half court. Bellamy despite a significant minutes reduction from 1966, increased his productivity by almost 20% following the 1968 season.

Seven of the nine 1966 NBA franchises improved productivity at the Center position by 1969. The Bullets with Unseld/Ellis increased production value by 106%. The Lakers, with the addition of Chamberlain in 1969, increased productivity by 195% from the season ending 1966. The Sixers traded Chamberlain for three players including Center Darrall Imhoff. Imhoff, was the second center on the Lakers depth chart in 1966, playing behind Leroy Ellis. Imhoff, although an average player, proved to be better than Ellis as a starter.

The seven teams that outperformed their 1966 output at the Center position (Knicks, Royals, Bullets, Warriors, Hawks, Lakers, Pistons), didn't just improve by beating up on the expansion teams. The three expansion teams inducted between season ending 1967 and 1968 (Bulls, Sonics, and Rockets) had rebounded to produce Centers that performed over the 1966 average levels.

On an all-time scale, average productivity levels in 1969 had increased over 1966 in every position, despite expansion. The NBA was a league growing exponentially in popularity and resources. Imagine if the NBA could've retained star players like Rick Barry, and drafted players like Connie Hawkins, Julius Erving, Roger Brown, Mel Daniels, and Artis Gilmore, the early 70s could've been the golden era.

Increased productivity indicates dilution of competition.

If all the centers in the league averaged 13 rebounds per 36 minutes in 1970 -- and these same players average 14 Reb/36 in 1971 -- it's unlikely they all (on avg, as a group) suddenly became better rebounders. Most likely, it's because the league expanded from 14 teams to 17, and a bunch of bench-warmers are suddenly playing big minutes.

Mike G wrote: ↑Sun Jul 05, 2020 12:01 pm Increased productivity indicates dilution of competition.

If all the centers in the league averaged 13 rebounds per 36 minutes in 1970 -- and these same players average 14 Reb/36 in 1971 -- it's unlikely they all (on avg, as a group) suddenly became better rebounders. Most likely, it's because the league expanded from 14 teams to 17, and a bunch of bench-warmers are suddenly playing big minutes.

I use my own statistical model to assess every single NBA/ABA season from 1950 - current, but win shares and other stat models echo this in their measure of player productivity -- the league quality increased in the late sixties/early seventies despite expansion.

According to win shares Russell, Bellamy, and Chamberlain show some decline over the three seasons following season ending 1966. Now, this may be more so or even entirely due to age and attrition as opposed to the increase in league quality, but in comparison to the average productivity levels for the Center position in 1966 (1966 Teams averaged .116 W/S 48 from the Center position) Russell, Bellamy, and Chamberlain were all performing well above the 1966 average.

The argument that Russell, Bellamy, Chamberlain along with Beaty and Thurmond -- who showed no decline over those three seasons -- simply feasted on a weaker quality of competition doesn't seem supported by the value metrics. By 1969, the five expansion teams had produced two Centers in Rule and Hayes that would've been top five Centers in 1966. The Knicks and Bullets increased their Center production by 44% and 107% respectively, due largely to the emergence of Reed and Unseld.

In the next year, season ending 1970, although Chamberlain missed most of the season due to injury, Russell retired, and Beaty singed with the ABA, Abdul-Jabbar debuted while Hayes, Unseld, and Reed continued their stretch of dominance. The middle-of-the-road Centers like Boerwinkle/Imhoff/Fox were better than the average Centers of 1966 -- Ellis/Strawder/Embry. By all accounts, the 1970s competition quality was better across the board than in the mid-60s, but nowhere was this measure more definitive than at the Center position.

My value shares formula does a much better job at drawing these comparisons because it's designed to compare performances from different periods and different seasons. I'm using win shares simply because it's the more recognizable formula.

There were 14 teams in the 1969-70 NBA (call it 1970), and 3 expansion teams arrived in 1971.
The 14 teams that averaged 0.0 (naturally) in point differential in 1970 would avg +2.0 in 1971


This was in spite of losing some players in the expansion draft. Why would teams so weakened tend to do better?
When you look at players on the 3 expansion teams, they almost entirely got more minutes and likely improved their per-minute numbers even more: The schedules gave them 12 games against each of the other 2 new teams (and just 4 matchups against established teams).

Several factors at play here:

• Size of Talent Pool
• Ability to obtain elite talent from pool
• Rules of competition
• Number of players in League

Remember, the talent distribution in the league approximates the upper tail of a bell curve, where the bell curve represents all people in the talent pool.

What I found (years ago, when I analyzed this) is that you can "fit" a talent pool size to the shape of the curve from league average up to elite players. In other words, assume all of the players in the talent pool above NBA league average are in the league.

I posted a chart like this on Twitter a few years ago:

This is similar to what Tango understood for baseball years before this (2004?). http://tangotiger.net/talent.html

I am saying this as a framework for assessing NBA talent levels.

Here's a thread where I attempted to implement the assessment of league talent levels/ quality of competition, using the "delta method" originated by Neil Paine: http://apbr.org/metrics/viewtopic.php?p=16097#p16097

Mike G wrote: ↑Sun Jul 05, 2020 8:12 pm There were 14 teams in the 1969-70 NBA (call it 1970), and 3 expansion teams arrived in 1971.
The 14 teams that averaged 0.0 (naturally) in point differential in 1970 would avg +2.0 in 1971


This was in spite of losing some players in the expansion draft. Why would teams so weakened tend to do better?
When you look at players on the 3 expansion teams, they almost entirely got more minutes and likely improved their per-minute numbers even more: The schedules gave them 12 games against each of the other 2 new teams (and just 4 matchups against established teams).

I'm not sure if your method gives the best assessment of league quality. That method would assume that there's no change to league quality without expansion.

The NBA expanded to 22 teams in season ending 1977. The 18 NBA franchises from 1975-76 saw "no" change in their net efficiency ratings (the more advanced way to measure team performance differentials) in 1976-77 despite adding "four" new teams.

There are variables that your method of simply using established team point differential or (efficiency differential) don't account for or explain: The relative strength of incoming draft classes, the unexpected contributions from player transitions (eg. initially underrated players like Bob Love, Bob Kauffman come to mind), and coaching/schematic changes -- as I mentioned in the OP, Holzman transformed the Knicks from the worst defensive franchise in season ending 1967 to the best in the NBA by 1969. These events are reflected in the team differentials but are not a function of established teams beating up on the expansion teams.

Another point about draft classes to remember when assessing the competition quality--take into account the existing strength of the league at the time. The 1959 and 1960 draft introduced in all-time players like Chamberlain, Robertson, and West into a developing league with few other superstars matching their immediate value. In the 1968 draft, Hayes and Unseld were coming into a league where Chamberlain, Robertson, and West were still in their primes. The late sixties and early seventies average Centers like Boerwinkle, Fox, Counts were better than average Centers like Barnes and Strawder from a few seasons before. I think these are better indicators of advancements in competition quality than team differentials for the aforementioned reasons.

The NBA expanded to 22 teams in season ending 1977. The 18 NBA franchises from 1975-76 saw "no" change in their net efficiency ratings in 1976-77 despite adding "four" new teams.

Actually you are talking about the merger with the ABA, in which 3 ABA teams were dissolved and 4 intact teams entered the NBA. This represents a contraction from 25 teams to 22, and in fact the avg NBA player saw reduced production. In particular, the ABA dispersal brought some premier rebounders: Moses, Gilmore, Lucas, Kenon, and Nater -- 5 of the top 10 rebounders in 1977 were gifted from the ABA.
EDIT -- Kenon played for SAS in both leagues. The others became top rebounders for their new teams.

Mike G wrote: ↑Mon Jul 06, 2020 2:42 pm

The NBA expanded to 22 teams in season ending 1977. The 18 NBA franchises from 1975-76 saw "no" change in their net efficiency ratings in 1976-77 despite adding "four" new teams.

Actually you are talking about the merger with the ABA, in which 3 ABA teams were dissolved and 4 intact teams entered the NBA. This represents a contraction from 25 teams to 22, and in fact the avg NBA player saw reduced production. In particular, the ABA dispersal brought some premier rebounders: Moses, Gilmore, Lucas, Kenon, and Nater -- 5 of the top 10 rebounders in 1977 were gifted from the ABA.
EDIT -- Kenon played for SAS in both leagues. The others became top rebounders for their new teams.

C'mon.

The Kentucky Colonels were never an NBA team for instance. The NBA never "contracted" in season end 1977, it was an expansion through the merger. There were 18 teams in season end 1976, 22 by season end 1977.

The only difference is the NBA populated the four expansion teams in 1976-77 with a 75% carryover from various 1975-76 ABA rosters, as opposed to the three 1970-71 expansion teams that filled 40-45% of their roster spots through the 1970 draft. The late 60s/early 70s NBA drafts were expanded to the extent that the 12-14 teams were able to sign from 230+ selections--All-Star Randy Smith was notably selected by Buffalo in the 14th round I believe. The roster spots vacated by the unprotected players in the 1970 expansion draft were filled by rookies from a relatively deep draft class.

The primary reason, I argue, that the NBA was able to raise competitive quality in spite of expansion and ABA competition, is the soaring popularity of the league starting from the early 60s Russell/Chamberlain marketing campaign. The unprecedented interest brought about additional resources and better recruits--as a resulting in a more quality product.