There’s no I in team but is there a team in baseball?
Does switching team affect baseball player's performance?
Baseball is in-theory a team sport. Two starting teams of 9-10 players (depending on which league you play in), with squads of 25 players face off against each other. Fans are loyal to teams rather than just players. The New York Yankees is the one of the well-known sports franchises in the world.
But are they teams or just a collection of individuals? Baseball is a series of 1-on-1 matchups between pitchers and batters, interspersed with brief moments where the 9 fielders need to coordinate. We can measure how well each player does in these interactions. Does the batter hit a homerun, or the pitcher strike him out?
In my previous blog, I showed that basketball players’ performance depends on their team. Basketball players who switch teams were more likely to perform much better or much worse, than players who stayed at the same team. This showed that in the NBA, its not individual player ability but also team chemistry that matters. This makes sense as basketball players need to coordinate in both offense e.g. passing and defense e.g. marking.
In this blog I test a similar theory for baseball. Does changing teams affect a baseball player’s performance (as it does in basketball) or is a player’s performance purely a function of their own ability? Given so much of baseball is 1-on-1 interactions, my theory is that teams will not affect baseball players’ performance.
I look at the year-on-year change in performance and compare traded players to players who remain at the same team. The below graphs show Wins Above Replacement (WAR) for MLB batters and then pitchers in consecutive seasons. WAR measures a player's overall by deciphering how many more wins he is worth than a replacement-level player at his same position (e.g., a Minor League replacement or a readily available fill-in free agent). I include all players between 2000 and 2019, who play 2 consecutive seasons between 2000 and 2019 with either 20 Innings Pitched or 70 Plate Appearances.
There are two things that stand out. Firstly, for both pitchers and batters the average performance of players who switch teams is lower than for those that stay. The average transferred batter adds 0.5 wins to their team, while non-transferred batter add 1.2 expected wins. Teams are clearly less willing to transfer their more valuable players.
Secondly, there does not seem to be increased volatility for traded players for either pitchers or batters. If individual performance was affected by teams, players who switched teams would be more likely to see large improvements or setbacks in performance. We might expect traded players to be further from the dotted line, which indicates performing equally well in both seasons. It does not appear they are though.
To test things further, I run a regression on a range of performance metrics. I compare the absolute difference in performance for players who transfer against those who do not. I control for other factors that could confound the estimates: these are the number of plate appearances (batters)/ innings played (pitchers) age and individual fixed effects. More information on the methodology is available in my previous post.
As expected, switching teams does not seem to matter for overall performance of batters or pitchers. The bars show the estimated effect with 95% confidence intervals. There is no statistically significant effect on Wins Above Replacement or most of the other performance metrics.
There are two other interesting results though. DRAA (Defensive Runs Above Average), which measures fielding value, is slightly more volatile for players who switch teams. This is intuitive as fielding is one of the only times when baseball players need to coordinate with other players. Players who move to a team where their fielding style fits better/worse with the other players will perform better/worse.
Also, the amount batters strike out (K%) or get walked (BB%) is volatile following switching teams. This suggests that batters are more likely to change their style when switching teams, by swinging more or less than they did previously. This may be due to different coaching. Swinging more does not automatically improve performance, as there is a cost to swinging more (more strikeouts) as well as a benefit (more walks.) So switching teams affects how players perform at the margins, but the overall effect is pretty much nil.
Team construction still matters in baseball. You still need batters who can field every position and putting certain batters next to each other has been shown to be advantageous. However, unlike in nearly every other sport, we can separate individual performances and show how they sum to the whole performance. This does not detract from it at all. It’s what makes it so fun for stat obsessives.
Performance Metrics:
Data is taken from Neil Paine’s Advanced Metrics dataset and Jeff Lahman’s data.