Four years ago, I went to Thailand with my cousin where we rode mopeds, pet stoned tigers and watched Muay Thai. We also visited a Buddhist temple where I ran into a quote that I sometimes remember when I’m trying to decide how much I should care about something. It was:
“Remember, one hundred years from now, all new people.”
So the monks were clearly ballers, knowing that nothing, including NBA teams, lasts forever. In that spirit, let’s put on our GM hats and figure out how long a team lasts.
This retention chart may look daunting but actually isn’t so bad. Each row represents when a group of Mavericks started their tenure (and played in the regular season). The ensuing percentages to the right represent how many of those Mavs stayed on each subsequent year. For example, 38 percent of the Mavs who played in 1996-97 came back in 1997-98, then 29 percent came back in 1998-99, then 19 percent came back in 1999-00 and so. Go down a row and you’re looking at the Mavs who joined in 1997-98. This goes back 20 years, covering 174 Mavs total.
Right away, you see the BMW of the Mavs, Dirk Nowitzki, cutting through my chart like a one-legged fadeaway knife. But let’s look at this in detail.
- The gaps in the rows represent Mavs who left and then boomeranged back. The full list in chronological order: Greg Buckner, Juwan Howard, Eduardo Najera, Adrian Griffin, Devin Harris, J.J. Barea, Tyson Chandler. Looks like Donnie has a thing for lunch-pail veteran types. In fact, there’s only been one year in the last decade without a Boomerang Mav (2012), sort of reinforcing the notion that Dallas is a good place to play if the Front Office can get to them here in the first place.
- There’s no denying a sense of dissatisfaction as we cycle through one-year deals but the fact is the average Maverick cohort tenure has been pretty steady going back to the Clinton era. Yes, the last two years have been revolving-door-like but the fact is a minority of Mavs make it to three years and a majority don’t even get a second year, even before the 12:01 Meetings era. In fact, the average Maverick tenure is only 1.1 years and that’s boosted by Dirk.
- In general, the chart is kind of blotchy because the denominator is in the teens. Let’s grow that denominator by looking at the whole league and normalize the start years.
Ooh, colors! So, you see after three years, the average NBA roster is only keeping 24 percent of their team from four years ago so the Mavs aren’t THAT much of a turnstile, national sportswriters. Also, it means that history might bet on a Splash Family Divorce by 2020. If not, we are sending Adam Silver to spice things up.
Just to reiterate, if we look at the retention curves for the last 20 years, the Mavericks do quite well vs. the NBA (if not flipping your whole roster is good). This chart shows on average how many players last from their start year. So at five years, about 20 percent are still around for Dallas.
There’s Dirk, kicking his leg again! So, Dallas is actually fairly stable. Let’s look at the whole league now (focus on the left half of the chart when the years are under 10 or so as outliers like Dirk and Kobe start biasing the larger cohort years):
You see the curves are fairly consistent across the league. ATL, DET, IND, LAL and SAS pop a bit. Their retention curves hover in the low 20s all the way up to year 6 and year 7, which is remarkable given the average player tenure with a team in the NBA is only 1.7 years.
Let’s see if this team retention stuff matters when it comes to getting wins. I’m going to key on the third year retention due to first-round rookie contract guaranteed lengths and also because it’s a nice run for a head coach/GM/player nucleus. Let’s compare franchise winning percentages over the last 20 years to the team’s 3-year Player Retention Index. Each blue dot represents a NBA team and where its win-loss percentage and 3rd year PRI are.
Running a regression, it looks like roster churn has some mild influence on wins and losses. Without sounding like a jerk and saying “it’s correlation, not causation”*, it does seem as if teams that don’t make themselves over less tend to win more games, which feels right. Like cars that stay out of autoshop tend to run longer.
* this is the analytics community’s unofficial second favorite retort after “what’s the sample size?”
Anyways, I used to think it was kind of fun that all our bodies’ cells change over every 7-10 years (until Google told me it was wrong). Now, we can think about how 80 percent of a NBA team changes over every 3 years (until a self-driving car from Google hits me in the street and tells me I’m wrong).
Somethings will never change. Like change.
All stats courtesy basketball-reference.com.