At the end of a season you wonder – which of these players will be back next year? Some will move up to higher-level leagues. Some will move horizontally. And some will step down a level or out of the game completely.
For fans it’s a time of excitement.
For players it can be a time of anxiety.
For clubs, it’s a time of assessment and revision.
This article assesses roster turnover from 2019 to 2020, and 2020 to 2021 for each Canadian Premier League club. Some clubs changed little, while others blew up their rosters.
2019 to 2020 turnover
The Wanderers’ actual roster stability was even lower than the numbers show. Jan-Michael Williams played a lot of games as a Wanderers’ goalkeeper in 2019, and retired after the season to become the club’s goalkeeper coach. But he also signed an emergency player contract in 2020 due to injuries among the club’s active goalkeepers, so he counts in the analysis as a returning player even though he never played in 2020.
2020 to 2021 turnover
Roster stability rates from 2020 to 2021 were similar to those in the previous off-season. Pacific FC kept pretty much the whole 2020 team together, while York discarded not just their 2019 name but also many of their 2019 players.
I did this analysis in early July, so any players signed since then won’t be represented (e.g., Klaidi Cela didn’t re-sign with Forge until August, so he’s not included as a returning player). I meant to publish this blog post at the start of the season when it would’ve been most pertinent, but life gets hectic sometimes.
Is CPL roster turnover normal, compared to similar-level leagues?
After the 2020 Island Games, Donald Belcham published a thorough analysis (here) of roster turnover from 2019 to 2020 in the CPL. He didn’t just calculate roster turnover rates. He calculated roster turnover rates in several similar-caliber leagues internationally over a bunch of recent years to get an idea of how CPL turnover rates compare. What is a normal level of roster turnover for a league like this?
His answer was that CPL’s 2019-20 turnover was pretty normal. Forge and Cavalry were a bit unusual in how stable their rosters were, but overall the league’s turnover matched its peers. That conclusion remains the same for 2020-21 since roster turnover rates remained about the same.
Is roster turnover bad?
Not necessarily. Sometimes good players are sold on to higher leagues (e.g., Cavalry selling Joel Waterman to CF Montréal of MLS). That’s generally a good thing. Sometimes bad teams get rid of the players that made them bad – that’s generally a good thing too. But if teams routinely have high turnover rates it raises flags – why don’t players stick around with that club?
There are obvious benefits of roster stability – with team cohesion players are familiar with each other’s tendencies on the field, so it’s logical to expect improved team performance. Roster stability is a goal for some CPL clubs.
“I think consistency in a squad is key. And that’s really your foundation, your base—and then as players come and go, whether that be on-field reasons or off-field reasons, you’re going to have to always make changes, try to add quality on and off the pitch, but certainly, that [consistency] always should be the foundation of a club.”Rob Friend, Pacific FC CEO (source: this interview with Martin Bauman)
But how good is roster stability, for real?
Does roster turnover help or hinder clubs’ performance?
In my mind, I think of course roster stability helps performance. If you’ve played any level of soccer, you know that familiarity with teammates helps you connect for passes as you know each other’s tendencies. But this is an analytical article, so let’s look at the data.
The CPL is too young to detect any trends, so let’s look at two analyses of other leagues’ roster turnover, and its relationship with clubs’ performance the next season.
Analysis 1: other leagues similar to CPL caliber
As part of his exploration of CPL-caliber leagues’ turnover rates, Donald Belcham asked: was roster turnover related to a change in the team’s standings? You could reasonably expect any of three outcomes for teams that changed over their roster a lot in an off-season.
- They improved in the standings. Like if they get rid of under-performing players or sparked life in the dressing room or something.
- They declined in the standings. Like if the players are new to each other and have no cohesion, passing through-balls to players not making runs.
- They didn’t tend to improve or decline in the standings. Like if it’s complicated.
He found outcome #3: there was no relationship between teams’ level of off-season turnover and subsequent change in league standings. Roster turnover itself doesn’t help or hinder – it depends on which players are coming and going.
There have been a series of analyses of MLS teams’ turnover rates, with some of them relating turnover rates to subsequent team performance. I’ll lay out their main findings in the bullets below, and then discuss why I’m skeptical of some of those conclusions.
- Eliot McKinley calculates roster turnover every off-season, in terms of how many player-minutes from the previous season are coming back the next season. Based on his evaluations of 2018-19, 2019-20, and 2020-21, around 75% of player-minutes return from season to season.
- Analysing 2013-15 MLS data, Matt Bernhardt found the following (source analysis here):
- the players departing MLS clubs tended to be low-minutes players.
- teams that did well in the standings tended not to change their roster much for the following season.
- Some analyses suggest that MLS teams with little roster turnover tend to get more points than teams with high off-season turnover (source: this analysis by Eliot McKinley). This seems to be especially pronounced in the first five games of the season, when team cohesion/familiarity would be most evident compared to teams with a bunch of new players (source: this analysis by Kevin Minkus).
In sum, MLS teams that do badly in one season tend to change up their roster before the next season. And teams that remain stable year-to-year tend to do a bit better in the standings than high-turnover teams.
But I’m not convinced. Here’s why.
Spurious correlations (or how bad teams change but stay bad)
Teams with little off-season turnover tended to do a bit better than high-turnover teams (3rd main bullet above). The suggested explanation is low-turnover teams benefit from team cohesion. Makes sense. But what if there’s another reason? Based on the 2nd main bullet above, teams finishing high in the standings tend to have little turnover before the next season – they keep the gang together. So those teams with little off-season turnover doing well … it’s no wonder they tend to do well. Maybe it’s not because of team cohesion so much as that they’re usually good teams that kicked ass the previous season.
This could be a spurious correlation, where two seemingly-correlated variables are actually both being affected by a third variable. In this case, team quality (standings result from last season) may affect both turnover rates (good teams don’t change their roster much) and standings in the next season (good teams tend to stay good).
Maybe I’ve misinterpreted it. If so, let me know. There is some mis-match of years – some of those above-listed analyses are for 2013-15 while others are 2018-20. I would like to see a thorough analysis of MLS teams’ turnover rates and standings positions in the season before and after, with recent data. A general linear model could help tease apart the relative influence of roster turnover rates and the previous year’s standings on the next year’s standings. Maybe there’s even a significant interaction – e.g., maybe bad teams that change their roster rise in the standings while good teams that change their roster go down in the standings. Yes this is getting nerdy – it’s an analytics article so I assume you’re into that or what are you doing here.
The CPL is still a new league. Will rosters stabilize as the league matures? Or is the current level of roster turnover – a rate similar to international peers – what we can expect to continue? Time will tell.
Jay enjoys coaching and playing soccer recreationally. He’s dabbling in soccer analytics about the Canadian Premier League and League1 Ontario. He lives in Peterborough, Ontario with his family.