A few weeks ago I took a look at the performance of the league’s oldest players and how it compared to historical results. Now I’m back with more data!
While the data for the past ten seasons doesn’t have much of a pattern, when we include seasons dating back to 1978-79, there is definitely a more pronounced trend. In fact, now we can see that there was a golden age for the elder NBA player around 1996-97 to 2004-05. Interestingly, the two lock-out shortened seasons of 1998-99 and 2011-12 appear (for the moment) to be local maximums. Is there something about a lockout that makes older players more attractive to GMs? We’d need more data to know for sure, and although I hate lockouts, I’m sure we won’t have to wait all that long before we get that next data point.
This graph shows the minutes played by players aged 35+ by season, starting from 1978-79. Keep in mind that the lockout seasons, 1998-99 (21 on the graph) and 2011-12 (34 on the graph), are not corrected to reflect the decrease in total available minutes in those seasons; the same is also true of the current season (35 on the graph).
It’s clear from both these graphs that older players are playing a larger part in today’s NBA than they did during the 80s and early 90s. But to me the real question is: are they any more productive, and are this season’s older players playing better than we would expect?
But first: an interesting finding. When we check to see if there is a relationship between productivity and age, there is only a weak relationship, which is contrary to my previous finding. The graph above replicates the methodology from my previous post, but includes seasons from 1978-79 to 2012-13 instead of just the last ten years. However, there may be a problem with applying this methodology to older seasons; from 1978-79 to 1989-90, the average number of players aged 35+ was under five (in both 1980-81 and 1981-82, only Elvin Hayes made the cut). One or two players can have a significant impact on the average of such a small group. The solution is to ignore which season the players played in and to group them all together. When we do so, the relationship becomes even weaker (the R-squared is a minuscule 0.000445). Even if we add a filter for minutes played (400+, as always), the relationship remains weak (the R-squared is 0.00589) and the trendline is actually positive — which means that it actually has players becoming more productive past their 40s. Simply put, within the oldest NBA players, age does not do a good job of explaining productivity. This is probably due to survivorship bias and the fact that we are keeping track of WP48, not year-to-year decreases in individual WP48.
Despite this, it does seem that older players have been more productive in recent years.
In the past 14 seasons, older players have been consistently near the top of their historical range. While the ages have been within normal boundaries (odds of both populations being the same with respect to age: p = 0.237), the WP48s have been clearly different (odds of both populations being the same with respect to WP48: p = 0.00773). Furthermore, this difference has been driven by high productivity over the last five seasons.
Given this, I’d say that the league’s older players are indeed becoming more productive. Have they been exceptionally productive this season? Compared to the historical record, yes. But compared to the last five seasons, this season does not stand out. So it really depends on one’s perspective.