Let’s abandon the five-man rotation

The Cleveland Indians are coming off dominant outings by Corey Kluber and Carlos Carrasco, the top-end arms allowing just six hits in a combined 17 innings of work. But is this the recipe for long-term success? 

It has been almost 50 years since the MLB has been using a five-man starting pitcher rotation. Before the early 70’s it was normal for a team to go on a four-man rotation, and before that a three-man. But for almost five decades now it has been normal for starters to grab the ball every fifth day. Baseball is a game of tradition, and often times when tradition is broken, fans and front offices alike can get uncomfortable.

There’s a great chance using five pitchers instead of four raised some eyebrows. And before that I’m sure people hated the idea of four pitchers instead of three. And before that I’m sure people hated the idea of not having the same guy throwing every single day. And before that I’m sure people hated the idea of people throwing overhand. The before-that’s are endless. People fear change, but they eventually adapt to it, be it by choice or chance. So while you may not like the idea of ditching the five-man rotation, it is time to go back to either a three- or four- man rotation.

Below is a chart of the 100 pitchers who made at least 24 starts for their teams last season. With each team having five pitchers, each pitcher should start around 32 games. Twenty-four starts represents 75 percent of 32, and I decided I wanted to notate those who pitched at least 75 percent of their projected starts. Why did I pick 75 percent? Because I like numbers that end in fives and zeros. I just got lucky that it ended up being 100. Round number bias is a part of life, people.

What I did was take each pitchers stats against hitters for the first, second, and third times they faced them throughout a game. I also took their numbers for pitches 1-25, 26-50, 51-75, 76-100. I left out the fourth time through the line-up and 101-plus pitches because there were between 13,000-18,000 fewer at-bats for those categories.1

Time through lineup splits

I’ll be honest: I was very surprised with this data. I thought the numbers the third time through the lineup and after 75 pitches would be way worse than they are. However, I still think the improvement for hitters is significant enough to not let a pitcher go through the lineup three times.

As you can see, walk rate is pretty steady throughout an entire game, and strikeouts go down the more times a pitcher sees an opponent. The third time through the line-up pitchers generally give up a run. The spike in batting average and OBP is only 13 points, however slugging rises 80 points from the first time through compared to the third. So while the amount of runners that pitchers let on isn’t too significant, the amount of extra base hits and runs (likely the product of harder contact) they give up is.

I also found it interesting that batters perform best against pitchers for pitchers 1-25, when pitchers are at their freshest. I think you can attribute that to the fact that pitches 1-25 usually come against the top and middle of the order hitters, who should be the offense’s best players. Pitches 76-100 are when the offense begins to gain an edge, though, which makes sense considering it is normally the pitch count when they go through the order for a third time.

It’s clear that the longer a pitcher stays in, the more vulnerable he becomes to extra base hits and runs, as well as a decrease in strikeouts. While the numbers don’t look very significant over a small period of time, such as once through the line-up, overtime a difference can be seen.

If a starter pitches to the offense for a third time every single game, we can expect, based on the data above, 175 runs given up throughout a season. However, if teams pull the starter after two times through, and put in a new pitcher to go through the lineup for the first time, we can expect about 133 runs. In Major League Baseball we estimate that about 10 runs is equivalent to one win. So in this case we are looking at four wins throughout the entire season by replacing a starting pitcher with a quality bullpen arm.

The same goes for pitch count. For an entire season, pitchers will give up 172 runs during pitches 76-100. Compare that to pitches 1-25, when a pitcher is expected to give up 137 runs, that’s roughly three to four wins. Four wins is a pretty significant figure, one that gets larger when you smooth out the impact of the order-based impact discussed above.

It’s important to remember that not all innings are created equal. Would you rather see Corey Kluber start and pitch the first three or four innings, where the score has a higher propensity to be close, 50 times in a year, or would you rather him pitch six or more innings 30 times? While they may pitch the same amount of innings per year, your best guys are having impact on nearly twice as many games.

Obviously, you can easily flip that and say your worst starters are going to impact twice as many games, but you also have to remember that most pitchers are better their first time through the line up, and a lot of the damage gets done the third time through.

Take Josh Tomlin, for example. Here are his career splits:

Time through lineup splits Josh Tomlin

From the first to the third times though the lineup, Tomlin’s walk rate stays the same, but his strikeout rate drops considerably. Opposing batting average jumps nearly 40 points, slugging skyrockets, and total bases per plate appearance increases nearly eight full percentage points. Batting average on balls in play jumps (again, likely due to harder contact), home runs stay nearly the same despite considerably fewer plate appearaces, and tOPS+ (which uses each individual batter as his own benchmark) jumps 26 points. Alas, if you limit your “worst” guys to only two times through the line-up, you’re maximizing their talent.2

My rough draft goes like this:

You have three starting pitchers, three what I will call “piggy-back” pitchers, and then your six or seven relievers. In an ideal game, your starter will make it through the line-up two times in 4-plus innings. A perfect game would be six. Your piggy-backer will then come in to face the line up once or twice, hopefully carrying you into the seventh inning, where the bullpen will then finish the job the same way they do in baseball today. Obviously, we have to watch out for pitch count. Pitchers are generally pulled around 100 pitchers. Considering there is three starters now instead of five, they throw sixty percent of the pitches—the new number to look out for is 60. Managers will also have to manage the situation, as well. If the starting is on top of his game, you may want him to go out and face the top of the order one more time before bringing in the next pitcher. It would be no different than the decisions they make today, it would just come earlier in the game.

While this all makes sense mathematically, it would take time for a team and for players to ever buy into it. You’ve already seen elements of this in the postseason, teams riding their bullpen more often and in higher-leverage situations. Teams with a low budget are always looking for competitive edge’s to help them win. With starters commanding salaries that tend to reduce surplus value, this could be it. The Rockies tried a four-man rotation for a few months a couple years ago and limited pitchers to 75 pitches, but it wasn’t given a long-term opportunity to succeed. The numbers in the chart above are an average of MLB pitchers, but there are guys who get better as the game goes on. Some guys, like Trevor Bauer for example, tend to get roughed up their first time through the line-up—they need a few pitches to settle in. But if Trevor, a player who the Indians initially slotted in the bullpen a year ago, knew he’s only being asked to go four-to-six innings, perhaps his approach changes.

It may sound crazy because it is different, but one day, maybe just a few years from now, the three- or four-man rotation will be the new normal. Teams would be wise to adapt such sooner than later.

  1. Thank you baseball reference. []
  2. This is relative, of course. []