Baseball Musings
Baseball Musings
January 02, 2008
A Fifth of Starters

My latest column at SportingNews.com looks at the impact of fifth starters on teams. I was actually surprised at the high fraction of starts that go to the back of the rotation, mostly due to clubs being unable to field decent hurlers to take over due to injury. But pitchers you can classify as fifth starters on average get more starts than team aces.


Posted by David Pinto at 04:45 PM | Pitchers | TrackBack (0)
Comments

The Hardball Times did a series on ERA by rotation slot in 2006 (called "Fun with Numbers" I believe). Anyways, I remember coming away shocked when I saw the high ERA totals of 5th starters and how mediocre it was for a #3 starter. That is, when the Indians signed Byrd to be their #3/4 guy, little did I know how much better he was then league average.

Posted by: Brandon Heikoop at January 2, 2008 07:28 PM

Especially now that teams are reluctant to skip over the backs of their rotation when they have off days, the 5th starter is almost as important as the other guys.
I wonder really how long it will be until we see the first six man rotation.

Posted by: mlbgm at January 2, 2008 07:54 PM

Shouldn't an analysis like this determine fifth starters before looking at their stats? Besides the general idea that things like "ace" and "backend" are terms we usually use in terms of talent rather than post-hoc performance (or else there have to be 30 aces in baseball at all times), you're basically pre-selecting lots of starts, and moreover, lots of bad starts.

An ace can only make so many starts (for ease, let's say 35); meanwhile, the guy(s) with the fifth-most starts *plus* anyone else who made a poor -- dare I say it, replacement-level -- start. That alone explains the inflating ERAs of rotation slots post-hoc. But think about the starts. Obviously in a real-world scenario, you have to assume that it's near-impossible that an ace would make more starts than another rotation slot plus the majority of injury starts.

And as for the supposed poor quality of third, fourth, fifth, sixth+ starters, I think most people just look at the game unrealistically. Every team wants five quality starters, much less good guys to fill in; talent distributions and concepts like replacement mean that this is patently the stuff of dreams.

But the bottom line is that pre-defining pitchers as aces, fifth-starters, etc. surely seems like the way to go, no? And conflating fifth-starters with replacement-level guys just makes it worse.

Posted by: Dan at January 2, 2008 08:13 PM

While I agree with the overall premise, I challenge the counting of only starters who had ERAs higher than the #5 starter. This artificially inflates the group ERA. In the Mets example, if by some fluke (and in a sample of only 5 games it wouldn't have been that much of a fluke) if Pedro Martinez' ERA had been 6.50, he would have been counted as a 5th starter. Since it was only 2.57, he was not. Jason Schmidt is a good example of this. Based on his career numbers going into '07, he clearly would have been considered one of the top four Dodgers starters. However in six starts, his ERA was 6.31 and therefore was considered a fifth starter in this study.

I think a better definition for fifth starter would be any pitcher who:

1. Was not one of the first four pitchers used at the begining of the season, AND
2. Was not in the top four in games started for the team

You can find loopholes in this method too, but I would think that the biases would more likely even themselves out and have less of an impact on the overall average.

Posted by: Scott Segrin at January 2, 2008 10:07 PM

Scott and Dan,
You two are also missing that pitchers move slots during the season - through injury, match up, rest, or otherwise. The purpose of this article was to show how quickly the talent drops off after the 3rd and in some cases the 4th starter.

So using the Mets as an example, Sosa was presumably the #5 starter. The author does error in simply adding the following starters, however for sake of argument, we can assume that the Mets used Pelfrey+Sosa in the #5 slot (27 starts).

If that does not sit with you, we could then suggest that Pelfrey 'filled in' for 5 starts for Perez, and the remaining 8 of his starts for El Duqe. We still need two more starts to fill in the top 4 of the rotation, so we will take that from Lawrence.

But wait, that leaves the #5 starter with a total of 18 starts. Toss in Pedro, Vargas and the remainders, and voila, your #5. Probably worse then the author gave the team credit for.

Posted by: Brandon Heikoop at January 2, 2008 11:33 PM
Post a comment









Remember personal info?