Why is Shelby Miller in the bullpen?

Posted: October 16, 2013 in Commentators, Pitching, Post-season

I have Miller as way the better pitcher than Kelly – more than a half run per 9. Miller is projected very well by Steamer, had a very good FIP this season, and fantastic K and BB rates in his minor and major careers thus far.

Is there something wrong with him?

While we are on the subject of mediocre pitchers (Kelly), surely you want to pinch hit for him leading off the 5th inning in today’s game, down 3-2. Of course I want to get mediocre starting pitchers  out of the game as soon as possible, preferably after facing the order at most 2 times. Which brings up another interesting point: When a starter gets in trouble early and then settles down a bit, why is it important to still get him out of the game as soon as possible? Because he has likely burned through the order twice by virtue of getting in trouble! Folks – and I know I sound like a broken record – times through the order is everything for a starting pitcher. Not “how they are pitching” or pitch count, or anything else. Times. Through. The. Order.

You want a funny/ironic illustration of the nonsense that is spewed by commentators/ex-pitchers about how pitchers “are doing during the game?” After Greinke was in all kinds of trouble in the first inning, and then quickly retired the Cardinals pitcher (yes, the pitcher) to lead off the 2nd, Ron Darling, who I’ve lost all respect for (never really had much in the first place), remarked that “Greinke was now locked in after a shaky first.” What happened next? He gave up 4 straight hits and 2 runs!

And I’m not sure why Choate didn’t pitch to A-Gon and Ethier in the 8th. Isn’t that what he is there for (he is less than worthless versus RHB)? Gotta punish Mattingly for back to back lefties in the order.


  1. Mo says:

    Nothing to do with this, but do you have a twitter? Would be nice to know via Twitter when you post.

  2. TomC says:

    Well, one of your questions is easy. Choate came in to pitch to Gonzalez and Ethier in the 6th.

  3. TomC says:

    His spot led off top 8, but if they were going to PH Descalso top 8, they should have double-switched Descalso into the 9 and Siegrist into the 8 to begin with, and even after not doing that, PH’d for Kozma in the 9th and played a 2B out of position if the game continued. The entire sequence that forced them to endure a Kozma PA when they needed offense was really bad.

    • TomC says:

      Or at the very least, replaced Kozma with the pitcher instead of Descalso mid-8 if you had intention of letting Siegrist continue even if he could.

    • MGL says:

      Yup, if you are a Cardinal fan, having Kozma at the plate when you were looking for a HR (or any non-out) in the 9th was painful.

  4. Jim P says:

    Re: times through the order vs. pitch count. Would this imply that when a team faces a mediocre pitcher (or anyone but an ace, for that matter), they ought to swing early in the count instead of working the count to get to the bullpen? And on this note, does a batter who faces one or two pitches his first few times up have the same advantage on his third or fourth time through as a batter who has worked the count? I thought I saw a study that showed that there is no advantage to either batter or pitcher in long at-bats but am not sure.

    • MGL says:

      Very good questions! The first one. I don’t know about deliberately swinging early in the count. I think that there is an optimal approach for each batter and pitcher, but you are on the right track. The notion that it is good to try and get to the bullpen early (by running up pitch counts) with a mediocre starter on the mound is nonsense! The earlier you get to the pen, the better it is for the other team! (Again, notwithstanding and long term bullpen fatigue issues).

      I don’t know if it helps in future AB’s to have seen more pitches, but it makes sense that it might. That would make for a very interesting study. I will put it in my “things to do” list. Split the first time through the order into 2 groups. Those that saw 3 or less pitches and those that saw 4 or more (or some other split) and see how each group performed the second time through the order.

      As far as an advantage during long AB’s. I think that research was only for that AB, and not future AB’s. It would make some sense that if there is an advantage in future AB’s after a previous long AB that there should also be an advantage at the end of the long AB, but I don’t know. Sometimes looking for subtle effects is very difficult when working with sample data. It is too easy to make a Type I or Type II error.

  5. scapistron says:

    I know times through the order is the main driver of when to pull a starter, but has anyone looked at in-game kwERA or even just wERA? The idea being that walks and to a lesser extent strike outs are more indicative of how a pitcher is pitching than hits, home runs, or runs.

    Don’t want to use FIP because running into a homerun is a little random. Might want to exclude strikeouts because the batter tends to take an active part in that.

    Now that I’ve typed this out i’m thinking variation from their zone% could be even better.

    • MGL says:

      Certainly of you really wanted to see if there are any “hot and cold” hand effects within a game for a pitcher, the best way to approach it would be using pitch f/x. Conventional wisdom is that a pitcher may have his good stuff or not on any given day, and that that is persistent (i.e., how the stuff is in innings 1-X will tend to persist in inning X +1 and beyond).

      Absent that, FIP, or kwERA or wERA, or something like that is probably better than RA or even wOBA against. (I don’t know why you say that batters take an active part in K, but not in BB, by the way.) However, if you look at less subtle stats, like RA or wOBA against, you should STILL see an effect. It is just that the effect will be watered down somewhat and more prone to noise. But, as long as your sample size is large enough that should not make much difference.

      That is an important point in sabermetric research, by the way. You don’t always have to look at the most pure or rigorous stat to find an effect. If the effect and the sample size are large enough, that effect will show up even using very blunt stats.

  6. Brian says:

    This article –


    – does a good job explaining why Shelby Miller is in the bullpen. Might not be good reasons, but those are the reasons.

  7. MGL says:

    Yup. I did read that and forgot about it. Makes sense. That is a better reason than, “He hasn’t been pitching well lately…”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s