The Knicks seem to field almost an entirely new roster every season. But how do they compare to the league at large, and how much does roster turnover affect winning league-wide? New contributor Ted takes a deep dive into the effect the Knicks’ revolving door has had.

Since the end of the 2017-18 NBA season, the Knicks have had three different Presidents of Basketball Operations and four different head coaches. This, logically, has caused a large amount of turnover for the roster and in the rotation. Case in point, there’s now just one Knick currently on the roster who played for the team in the 2017-18 season, the much-debated and bedeviled French point guard, Frank Ntilikina. Every other player has since departed the Knicks roster — most recently Damyean Dotson, who signed with the Cavaliers over this offseason. Some have speculated how such high levels of roster turnover can be a source of struggle for the Knicks, and I wanted to investigate to see if this was true.

Since this is my first Strickland piece, I feel I should lay out the way that I think about basketball and analytics. I believe that analytics exist to, more often than not, disrupt the narratives and assumptions we construct when watching basketball. When watching a game, a season, or a team year over year, there’s too much data for us to take in, so our brains latch on to certain pieces of information and fill in things from there. This often leads to us coming to the wrong conclusions. An investigation using all the data available can help us reform those opinions, and in the future, hopefully come to correct conclusions more often. None of this is a new or unique philosophy, but I think it colors the rest of the piece.

I started with a very simple lens to analyze the Knicks roster turnover through — how many players were retained year over year, and how did the Knicks compare to the rest of the NBA? For simplicity, I decided to evaluate year over year roster turnover from the end of the 2017-18 season through the 2019-20 season — or two offseasons — in 2018 and 2019. In 2018-19, 11 Knicks that were on the 2017-18 roster returned to play minutes. That number fell to seven players from the 2018-19 roster to play for the 2019-20 team. Over those two seasons, the Knicks averaged nine retained players and just nineteen wins. NBA league average player retention during that span was nine players. Among the thirteen teams to average 41 wins or better over the past two seasons, average player retention was nine-and-a-half.

So case closed, the Knicks’ player retention has been average, our assumptions were wrong, and their futility has little to do with the roster turnover, right? Not quite. I found the previous approach to be imprecise and ignorant of context — is retaining a two-way player who played under five hundred minutes the same as retaining your starting point guard? What about a situation like Kevin Knox’s, who saw his minutes decreased by ten minutes per game, and after starting 57 of his 75 games played in his rookie season, started just four in his sophomore campaign? My analysis couldn’t account for these situations.

Serendipitously, while reading Dean Oliver’s Basketball on Paper, I found a statistic that does account for the context of minutes played — Roster Continuity Percent, which tracks the percentage of minutes played for a team played by players who were on the previous season’s roster. This gives the context that simply counting year-over-year players retained does not — retaining Kadeem Allen for just over 100 minutes isn’t counted the same as returning your 30 minute per game starters.

Roster continuity percentage is not as kind to the Knicks as my less precise method was, with 49% of the minutes played in 2018-19 played by retained players, and just 35% in 2019-20. For context, league average over that stretch was 62%. This reveals, as I suspected, that the majority of the players retained did not play major roles for the team the following seasons. While the Knicks may have retained seven players over the 2019 offseason, just three ended up playing significant minutes the following season — Mitchell Robinson (1400 minutes), Ntilikina (1200 minutes), and Knox (1150 minutes). Four of the five minutes leaders for the 2019-20 squad were added over the 2019 offseason. This means that while the Knicks may have brought back an average number of players every year, those players were small minute players, or had their roles shifted far more than they would on other NBA teams, as the Knicks cycled through players. This constant cycle creates an environment unideal for the development of young players — case in point, the treatment of both Knox and Ntilikina by the Knicks, where their yearly minutes, roles, and statuses in the organization change. Ntilikina has been a part of three Knicks administrations, with as many coaches as years in the league. Knox, as mentioned before, went from the Knicks’ number one draft pick to someone who could barely get off the bench at times. Their resultant struggles have been apparent.

To expand our area of analysis — between the 2010-11 and the 2019-20 NBA seasons, there were 22 seasons in which a team had a 40% or lower roster continuity percentage — five of those seasons were Knicks seasons, four of which have come in the past six seasons. The most seasons with lower than 40% roster continuity that another team had was two. Most of the teams that had a season under 40% were teams that saw their contention window open or close in that season — for instance, the 2014-15 Cavaliers, the 2012-13 Magic, or the 2019-20 Warriors. But the Knicks roster spontaneity has been based in futility and its consequences — new front office brings in their own players, players play poorly, fire front office; rinse repeat for the Knicks for most of the past decade.

Between the 2010-11 and 2019-20 seasons, league average roster continuity was 64%. Below is a chart with every team’s roster continuity percent for each season between 2010-11 and 2019-20. Seasons below league average are orange, seasons at or above league average are blue.

There are just two franchises to have just one season over the league average roster continuity percent — the Nets and the Knicks. Even teams like Minnesota, Phoenix, and Sacramento — fellow basement-dwellers of the NBA for most of the decade — have not matched the Knicks’ and Nets’ levels of roster turnover. The difference between the Knicks and the Nets this decade, outside of attendance numbers, is that the Nets landed Kevin Durant and Kyrie Irving in the 2019 offseason, just after posting their best record of the previous four seasons. That same season, Brooklyn posted a nice 69% roster continuity. It quickly fell back below average as a result of the moves necessary to acquire Durant and Irving.

The Knicks, on the other hand, will, for now, have to continue to toil through futility as they have for the past five seasons. And while the Knicks’ sole season above league average roster continuity, at 74%, was a disappointing 2013-14 campaign that netted just 37 wins, it’s no coincidence the Knicks’ second-best continuity percentage (54%) coincided with their best record of the century, in 2012-13.

So we’ve found that at least one bad team constantly turns over their roster, but is that trend true across the whole NBA? Do successful teams retain their players, while bad teams cycle through different players? To find out, I found the correlation coefficient (which is a statistic used to determine how strong a relationship is between two variables) between the average wins of each franchise and their average roster continuity percentage between the 2010-11 and 2019-20 seasons. The correlation coefficient, or r value, was .5775, which indicates a positive relationship between player retention and winning. I graphed all 30 teams’ roster continuity and average wins, which can be seen below.

Additionally, some of the best teams of the decade eclipsed the 90% roster continuity mark: the 18-19 Celtics (96%, 49 wins), the 18-19 Nuggets (98%, 54 wins), the 15-16 Warriors (95%, 73 wins), the 18-19 Jazz (92%, 50 wins), the 17-18 Blazers (93%, 49 wins), the Heat in 2013-14 (91%, 54 wins), the Spurs in 2012-13 (95%, 58 wins) and in 2014-15 (98%, 55 wins), and the Thunder, doing it three times, in 2010-11 (91%, 55 wins), 2011-12 (93%, 47 wins), and 2015-16 (93%, 55 wins). Playoff teams had a slightly higher than average roster continuity, at 69%, whereas non-playoff teams had lower than average roster continuity at 57%. Below is the same table as above, but this time, playoff teams are blue, and non-playoff teams are orange.

However, good teams are not good purely because they retain their players, but because they acquire players worth retaining. The Knicks could hold on to every player currently on the roster for the next five seasons without adding a single new player, but it wouldn’t necessarily make them a good team. More important than just retaining players is acquiring players who make you better and then retaining them. Part of that, however, is holding on to your draft picks — something the Knicks have done only since the Phil Jackson era — and drafting players worth holding on to, something the Knicks have not done. Simply put, retaining LeBron James wins you games, and not retaining LeBron James likely leads to losing, but retaining or losing Kevin Knox doesn’t move the needle as much.

If Leon Rose and the Knicks’ braintrust are serious about building a good Knicks team, they need to get roster continuity. While the likes of Julius Randle and Elfrid Payton are all likely on their way out after this season, the Knicks need to develop and retain the young talent they’ve already drafted (like signing Ntilikina and Robinson to extensions this offseason). Perhaps more importantly, whoever they select in this draft class needs to come into a stable team, where upwards of 60% of the rotation isn’t being overturned every year. If not, the likelihood that Rose will join Jackson and Steve Mills as former Knicks POBOs, after a disappointing stint, is high, and the roster turnover cycle will start itself over.