When I was a kid, I can remember looking at a basketball encyclopedia and being amazed by some of the statistics from the 1960s:

- Wilt Chamberlain once scored over 50 points per game.
- Bill Russell annually grabbed over 20 rebounds per game.
- Oscar Robertson averaged a triple double for an entire season.

And the list goes on.

I was born in the 1970s, but the decade I most closely identify with my childhood is the 1980s. We had Magic Johnson and Larry Bird and later Michael Jordan, but I can remember thinking “These guys are good, but they must not be *that* good. Their numbers aren’t close to those put up by guys like Chamberlain, Russell, and Robertson.”

What I failed to realize then — but realized later on when I started to look at things with a more discerning eye — was this:

- Players from the 1960s were playing in an environment where points and rebounds occurred with much more frequency than they do today.
- The talent gap in the 1960s was much wider than it is today, allowing for more extreme performances.

Let’s start with the first point.

Back in 1960-61, Elgin Baylor put up numbers that, without the proper context, are ridiculous: 34.8 points and 19.8 rebounds per game.

How ridiculous? Only one other player in NBA history has averaged a 34/19 line for an entire season.*

** That player, Wilt Chamberlain, did it the first six seasons of his career (1959-60 through 1964-65).*

But in the 1960-61 season, the average NBA team took a whopping 109.4 shots per game, a number that far outpaces the current average of 82.0 shots per game.

Although there are far more rigorous ways to do this, let’s take the simplest approach and reduce Baylor’s per game averages by 25 percent (a reduction that is proportional to the change in the average number of shots per game):

- 34.8 × 0.75 = 26.1 points per game
- 19.8 × 0.75 = 14.9 rebounds per game

Those numbers are still outstanding, but they also reveal that Baylor’s 1960-61 season is much more comparable to Kevin Love’s 2013-14 season (26.5/12.7) than would appear at first glance.

In fact, if we do this basic adjustment for every season of Baylor’s career, his career averages of 27.4 points and 13.5 rebounds per game drop to 21.8 and 10.8, respectively.

Once again those are still outstanding figures, but they also show that Baylor’s per game production was much more similar to Tim Duncan’s per game production (19.9/11.1) than the raw numbers suggest.

Now let’s move on to the second point, the talent gap.

In general, extreme observations are more likely to occur as the dispersion around the mean increases.

For example, suppose you have two groups of women. Both groups have the same mean height (64 inches), but group A has a standard deviation of 3 inches while group B has a standard deviation of 6 inches. Which group is more likely to have to have at least one six-footer?

Since the means of the two groups are the same, the group with the larger standard deviation is going to have more extreme heights, so in this case the answer is group B.

What does this have to do with the NBA? The plot below shows how the five-year moving average of the standard deviation of win shares per 48 minutes (WS/48) has changed over time:

As you can see, variability was much higher in the 1950s and early 1960s than it was at any other point in NBA history.*

** It should be noted that mean WS/48 is always right around .100.*

There are some other interesting features to note on this plot:

- The moving average took a sharp jump from the late 1960s to the early 1970s. This was due in large part to the NBA adding seven expansion teams from 1966-67 to 1970-71.
- The moving average significantly dropped in the mid-1970s. That decline was most likely due to the washing out of the expansion effect as well as the ABA/NBA merger.
- The moving average jumped again in the late 1980s and early 1990s. This was once again due to expansion, as the NBA added two teams in both the 1988-89 and 1989-90 seasons.

Getting back to the main point, some of the video game numbers that were posted in the early part of the 1960s were due in part to the large variability in player performance at that time.

Don’t believe me? In the table below, I present the number of seasons where a player logged at least 1,750 minutes and the percentage of those players who had at least 10 wins shares by decade:

Decade | No. | WS ≥ 10 |
---|---|---|

1950s | 339 | 13.57% |

1960s | 531 | 14.88% |

1970s | 1020 | 10.78% |

1980s | 1278 | 9.86% |

1990s | 1370 | 10.58% |

2000s | 1544 | 10.75% |

In a nutshell, seasons of 10 or more win shares were about 40 percent more common in the 1960s than they were in the decades that followed.

None of this is meant to denigrate the accomplishments of players like Chamberlain, Russell, and Robertson, surely three of the ten greatest players in NBA history.

Rather, the intent is to illustrate that some of the air needs to be taken out of those gaudy statistics from the late 1950s and early 1960s.

We don’t need to pine for the good old days. The superstars of today — James, Durant, Paul, Duncan, and Nowitzki, to name just five — are more than living up to the accomplishments of their predecessors.

by
Hi Justin – just a thought, but the standard deviation seems to drop right around the time the league started to become more integrated, going from essentially 4 black players maximum per team to something much more equal (obviously it has shot way past this now). Could the talent gap (i.e. high standard deviation in WS/48) be due to many NBA-caliber black players being overlooked for inferior white players?

That’s a good thought, and a distinct possibility.

Imagine how dominant Shaq would have been if you randomly took out 2/3rds of the black players from the game, plus half the 7 footers of either race? Now instead of Duncan/Robinson guarding him in SA, it’s Danny Ferry. Now instead of Divac guarding him in Sacramento, it’s Stojakovic.

Also numbers are changed by things like the 3-point line, or offensive goaltending rules (Wilt got a lot of his points & boards by tipping shots on the rim or redirecting teammate’s misses while they were on the way down and above the cylinder).

1962 Wilt got 26 rebounds per game (35% of his team’s total), while 1972 Kareem got “only” 17 rebounds per game (32% of his team’s total). So the disparity looks big, but actually is not.

I would like to know how many minutes a game had during Chamberlain’s era. More than 48 for sure if he played 48.5 a game once.

His average is greater than 48 minutes because the Warriors played in 10 overtime periods that season (48 + ((10 * 5) / 80) = 48.625 minutes per game).