Best of BP: The Gig Pitcher

We live in the era of McKinsey League Baseball. MLB and its teams are run mostly by people who have MBAs. So, when I say that we have reached the land of the baseball gig economy, you probably shouldn’t act so surprised. This is Baseball Prospectus. We… may have had something to do with that over the years.

We are beginning to see the beginning of the modularization of the baseball workforce. A few weeks ago, I looked at how teams are using the idea of “streaming” for their relievers. We live in an era where there are more good relievers than ever, and at the back end of the roster, teams are using wrinkles in the rules, including shuttling players up and down to the minors and using the injured list to—cover your ears Billy Eppler—swap players in and out. That way, there are more fresh relievers on the roster, even if they aren’t the same people on the roster from night to night.

But it hasn’t just been the relievers who have been multiplying. In the past decade, teams have been sending more players out start games over the course of a season than they used to.

Some of that is the advent of The Opener, but there’s another force lurking in the background. When we talked about relievers, we saw that teams seemed to have their favored high-leverage substitute pitchers who stayed on the roster, but there was a rotating cast at the back end of the pen. The innings that those pitchers get aren’t as important, and while the top end of any talent distribution is going to be set apart, as we get into the sixth- and seventh-best relievers on the team, are they really all that much better than the eighth-best or the best couple of relievers in Triple-A? Probably not.

Which brings us to the fourth and fifth starters in a rotation. Do they suffer the same fate?

Warning! Gory Mathematical Details Ahead!

Let’s return to the bullpen for a moment. The modern roster is loaded with relievers, and the average starter now logs just over 15 outs per game. This isn’t even an artifact of The Opener. We can look instead at the pitcher who recorded the most outs in the game and see the same basic effect. We know what’s happening here. Teams have recognized that when a pitcher gets to the third time through the lineup, their performance suffers. Whether that’s the result of familiarity with the pitcher or simple fatigue is hard to tell, but the effect is certainly there. With so many arms in the bullpen, teams have designed their pitching strategies so that especially the lesser starters don’t see the third time through as often.

We live in the world of Tutos and Threetos. Some pitchers go twice(ish) through the order. Some are allowed a third time around. It makes sense. Why ask a sub-par starter to do something that you know isn’t likely to work out well, especially when you have a bunch of relievers who can do the job better?

But once you’ve committed to that idea, there are some interesting things that follow from it that we find in the data. They include this graph:

I used the change in run expectancy for each event that happened for pitchers, and coded whether the plate appearance was on the first, second, or third time around the order. I then looked at each pitcher’s average performance on the measure each time through the lineup. The graph shows the standard deviation over the years among pitchers with a minimum of 100 batters faced in their role as starters in each category.

We can see in the graph that over time, there has consistently been less variance among pitchers their first and second times through the order. It’s not “everyone’s the same the first couple of times through the lineup” but the differences are smaller. It’s in the third time through the order that we see much more variability. When we look deeper, we see that one of the things that separates out the good starters from the umm… satisfactory ones is the ability to hold good performance through the third time in the lineup. We know that as a well-accepted truth in player development (pitchers who don’t have a varied enough arsenal won’t have enough tricks to turn a lineup over three times) but the numbers show it too.

Culturally, we’re still used to a baseball world where teams are fighting over the few pitchers who can hold that performance through the third time, because that’s what all five starters were expected to do. What happens though when you re-define the role of a couple of those starters – the umm… satisfactory ones—to only go for two(ish) times around the order?

Let’s slice our pitchers into groups. For each year, I took the first 30 pitchers in terms of outs recorded (before 1998, when there were only 28 teams, I took the top 28) and made that “Group 1.” These are our aces. I took the next 30 (or 28, as appropriate) as Group 2, and on down the line until I had five groups of pitchers. I looked at the average change in run expectancy for each of the five groups during the first time through the order. Everyone else was in the sixth group.

And then I did the same for the second time through the order.

We can see that the blue line, representing our aces, is the lowest—and here low is good—and more importantly very distinct from the other lines. The green line representing our no. 2s is above the aces, but also distinct from the gold, purple, and red mess above it. There are clearly layers, but they stick together more tightly than you might imagine. Again, it’s a little much to say that the first two times through the order, your number three and number five starters are essentially the same, but there’s also not all that much difference among them.

One of the things that we found with the development of the single-inning short-burst reliever was that short-burst relievers were better than starters on a batter-to-batter basis, although they burned out pretty quickly. But the fact that MLB hasn’t expanded in more than a quarter century, combined with an increase in the baseball talent pool from the further internationalization of the game and the fact that teams began actively developing short-burst relievers, has meant that there’s a talent glut in the bullpen. Sure, you have your seven or eight best relievers, but the space between the eighth and ninth best reliever in your system wasn’t all that large, and depending on who was rested and what matchups were coming up, it might make sense to flip flop them. If you have a group of relievers who are all somewhat indistinguishable from each other, why bother distinguishing them?

This is one of the hallmarks of the gig economy. There are some tasks where the result that you get isn’t going to vary all that much depending on whom you hire. There’s a plentiful workforce that has the skills necessary to do the work. I’m sure that someone out there is technically the best meal delivery driver possible, but really either way I just want my food, I don’t care who does the work, and it’s not like the difference between the best and worst is that big a gulf.

By redefining some of their starters to be short-bursters, teams have re-defined the position in a way that there’s a larger, more undifferentiated pool of players to fill it. Some starters who “failed” in a certain way became relievers. There’s a good possibility that some traditional starters who “fail” in a different way will become the next Tutos. The same talent pool expansion that filled the league with relievers could end up filling the league with Tutos.

We’re already seeing the result with relievers. Teams are streaming them onto and off the big-league roster, and MLB has already seen an increase in the disposability of its players. Player careers are shorter on average than they used to be, and tellingly, a lower percentage of players make it to arbitration and free agency, when their contracts start getting more expensive. Like everything else in the Efficiency Era, it makes sense and feels icky. Why pay someone $2 million to do a job that someone else can do just as well for $700,000? There’s no logical reason not to and so teams will do it, whether you like it or not.

There’s a decent case to be made that redefining some starting pitching assignments is a good idea from a gameplay perspective. The short burst reliever really is a better idea than a tiring starter, especially one who isn’t all that good to begin with. But in a world where MLB is run by MBAs, it makes business sense too. MLB can start to run some of its roster spots on the gig model.