A not-very predictive but interesting review of what player utilization might say about the quality of a team
Close watchers of the Cleveland Guardian in 2024 will note the carousel of players rotating through the lineup. I was thinking about this and thought it odd that despite this the Guardians sit atop their division.
I realized that in my experience teams that rely on a lot of different players are suffering from one of the following:
Thinking about this more I recognized that my expectation is that a baseball team that uses more players is more likely to stink than a team that does not. My reckoning is that even a team using an excellent platoon system is not likely to go through many, many players; good baseball teams usually have established stars who are expected to log 600+ plate appearances. From there, teams that have had to plug the lineup are typically filling in with less talented players. It is also possible for a team to be so poorly run that folks just keep trying different things and players and ideas and despite being in Denver everybody seems to be having a bad time. Alas.
There are other reasons that a team can decide to play a lot of different players, but these seemed to be the most common to me this morning.
The Guardians are, at least at the plate, not following any of these three tracks. My expectation before looking at the numbers was that the Guardian may very well lead the league in this invented statistical category: Players with at least 100 PAs as of the morning of September 15th, 2024 (Pw/100Pa9/15/24)
If forty to sixty hours of your life take place in Excel already and you’re not about to do that on a Sunday, here is a chart!
Let’s focus on that 0.5 X-axis line on the chart. That is the difference between a winning and a losing record and the most critical place to look. Something a little bit funny happens right there. there are only two teams in baseball with a winning record that have given at least 100 PAs to more than fifteen players. There is one team in baseball with a losing record that has used fewer than fifteen players. Fifteen is also the most common number of players to reach 100 PAs, with eight teams landing there.
A linear regression suggests that only 12.5% of a team’s record can reasonably be inferred by the number of players that they use. However, the White Sox might just be the worst team in the entire history of baseball, and so I present the regression again with that outlier removed. There is not much to see — the predictive power of the “model” barely improves.
This puts us in an interesting spot with the data set that we are choosing to analyze.
If you tell me how many different batters a team has used then it is not necessarily true that I can predict what their record is. This makes some amount of sense as at least half of the game is unaccounted for: pitching.
I think it would be interesting to look at this for an entire 26-man roster over the course of a season in addition to looking at just the predictive power of how many relievers, starters, and pitchers in total get used in a season. My hypothesis flying into that would be that needing to use many more starters than average tends to be more impactful than needing to use many more relievers.
Combining that in some way might lead to numbers that match a little bit closer to the expectation: a team that goes through a lot of players is probably not a very good team, unless they are doing something strategically interesting.
I will put a pin in this analysis for now. The most interesting thing is this: of the five best teams in baseball, four have used only 13 different batters for at least 100 PAs. The Guardians are that fifth team, and they have used 17 different batters for at least 100 PAs.
The other team to use more than fifteen players and log a winning record? The Twins.
I feel like something interesting is happening here that might require a broader scope and more data, but for now: The Guardians are doing it differently than the rest of the best.
