My friends on “Locked on Astros” have been talking about what is the best Astros bating order a lot recently. In 2019, I had investigated this topic and concluded that who batted in what order 1-7 really did not matter much that year. I had the same opinion about this year’s team.
Fortunately, I discovered a cool website that allows one to test different lineups. It uses a very simple model where the only thing one needs is the OBP and the SLG for each player. The website is here. You may want to play with the math model there. If you do let me know what you discover.
Given the data we need let’s lay out the Astros roster. Dusty Baker has had to use multiple lineups due to injury. I have tried to project what I THINK is his current standard lineup.
Some things one should instantly see in this team
This analysis excludes Siri, Gonzalez and Stubbs who all have too few PA to really analyze
The Astros have
Eight players above 0.800 OPS (120PA min)
Ten Players above 0.750 OPS (120PA min)
No player with an OPS greater than 0.905
That is an insane level of quality AND parity among the starters
Given this parity, I believe that as long as the catcher is batting ninth, Dusty Baker could almost randomly set the rest of batting order. Well, that is probably a little bit of an exaggeration.
What does the baseball musings model say about that?
For this baseline batting order the model predicts 5.523 runs per game. To help validate the model, the Astros have scored 5.340 runs per game this year and due to injuries and trades and promotions they did not use THIS lineup all year. As a side note, only the Tampa Bay Rays have more runs per game at 5.36. The model prediction of 5.523 runs per game seems completely reasonable for that lineup.
What if the lineup was the following, however?
Indeed, the baseball musings model says that lineup would be better and predicts 5.552 runs per game. The swap of Tucker and Brantley would add about 5 runs for the season.
Furthermore, the model predicts the following would be the optimal lineup.
For this lineup, 5.587 runs per game are predicted. This adds about 10 runs in a season over the base lineup. What would Astros Twitter say if Dusty started that lineup today? Hysteria.
The baseball musings model is very simplistic. What if one was able to incorporate style of batter vs. style of pitcher and park and weather effects on any night to determine the best lineup THAT night? I am confident that if we are not there yet, the teams will be soon. A manager can get a printout of who would be the best to use in what order tonight.
It is also kind of fun to consider who has been hot. Here are the same stats over the last 2 months (7/18-9/18)
What is the best lineup for team is not a static answer. The optimal lineup changes as players get hot and their health fluctuates. There have been several players that have either gotten very hot or cold over the last two months. If one knew this was going to happen the lineup would clearly be adjusted to take advantage of it.
Tucker and Bregman have been great at the plate. Jones in his part time in the last two months has the second highest OPS (excluding Siri’s limited PA) behind only Tucker. Brantley and Castro have been hurt and this has impacted their performance. Perhaps the time on the IL will be good for them in the postseason. McCormick has been challenged to keep up his hot start.
Overall, with the return of Bregman and Tucker playing like an MVP the past two months the previous lineups would be predicted to have performed better. There is also a different optimal lineup over the past 2 months. We said previously that the full year optimal lineup was 10 runs per season better than the baseline. Plugging in data from the last two months for the optimal lineup makes the model predict and additional 39 runs per season.
One thing is clear. Individual performance fluctuations are more important than the order when so many players are playing at near parity.
The official LarryTheGM lineup is the one labeled “Better” above. We will see how long it is until the Astros use THAT lineup.
Statistics from Fangraph.com