Adjusting the Lineup to Win

Every year when I get my Bill James Handbook, one of the first things I look at is the Manager Records. I enjoy seeing things like pitcher usage, tactics, and lineups used. As a Cardinal fan, we see more lineups a season than most other fanbases. In fact, Mike Matheny led the National League this past season with 146 different lineups, the most he’s used in one season. I also noted that it was his lowest win total and the only year he missed the playoffs.

Was this coincidence? My curiosity was piqued.

I pulled out a lot of my old handbooks to find out. I started by looking at the last 10 years of MLB to see what the year by year standards showed.

I placed all of the data in a scattergraph, using lineups for the X axis and wins for the Y axis. Looking at it year by year, there was no real trend to the data. Here was 2016:

But when you add a trend line, it shows you the correlation in the data. For 2016, the trend line shows a negative correlation; basically, the more MLB managers tinker with the lineup, the lower the amount of wins. Here is the data again, but with the trend line:

Looking at the previous seasons, the trend is pretty much the same; the data is scattered, but the trend lines are showing negative correlation. When all of the data is presented together, you get almost the same correlation:

Seeing almost the same results, I got to wondering about individual managers; does this trend mean all managers have a lower chance of winning the more they change their lineups?

I started with Matheny. In his five seasons, he’s used 122, 89, 199, 135, and 146 different lineups with 88, 97, 90, 100, and 86 wins. When plotted, they show a negative correlations:

5 seasons is too small of a sample size though. I expanded the research to 14 other managers who have had longer careers and varying degrees of success. I went with Dusty Baker, Bruce Bochy, Bobby Cox, Terry Francona, Joe Girardi, Clint Hurdle, Tony LaRussa, Jim Leyland, Joe Maddon, Charlie Manuel, Mike Scioscia, Buck Showalter, Joe Torre, and Ned Yost; all of these guys have been to the playoffs, but have also had some bad seasons. They all also have managed multiple teams, outside of Scioscia. The only restriction I used was they had to have managed a minimum of 140 games in seasons considered; some of them managed less than 100 games due to being hired/fired mid-season and those seasons would pull the data down. I also only used data from 1995 and on; finding lineup usage data before that is a challenge for a blogging hobbiest. Here is the graph with all 15 managers:

Francona and Leyland are the only managers to go against the trend; both have positive correlation between lineups and wins. I’d also say Charlie Manuel goes against the trend as his trend line is almost netrual; Scioscia is close to being neutral too.

Other things I’ve noticed:

• LaRussa is an interesting case. He had no seasons with double digits lineups used; in fact, his lowest total was 117. He aveaged 133 lineups a season and topped out at 153 (or a different lineup in 94% games managed). He also had 3 time where he used the same amount of lineups in back to back seasons: 120 in 1995 and 1996, 146 in 1997 and 1998, and 117 in 2001 and 2002.
  
• Baker’s highest win season just happened to coincide with his lowest lineup usage – 2000 with the Giants.
• Cox and Manuel might be the last of the old guard managers; they averaged the lowest amount of lineups at 94. In fact, they were the only managers to have double digit averages. I call them “old guard” because they always seemed to follow the old unwritten rules of the game and less inclined to sabermetrics.
• LaRussa was the highest averaged at 133. Maddon was second at 131.

My conclusion on this is there is an impact on lineup usage versus wins, but I believe it’s more about the reason for the different lineups. It’s quite possible that the increase in lineup useage is due to injury; a replacement player will impact the probability that a team will win. If I can get the data, I’ll write about lineup useage vs DL days per team to see if there’s correlation there.

I do believe that some of it is just tinkering. As a Cardinal fan, it’s obvious that LaRussa and Matheny adjusted the lineup daily; they both would play the matchups rather than setting a status quo in their batting order. It can be frustrating as fans, especially when their changes don’t bring a win.