Following Up on Some Questions

Lots of good comments in the post yesterday about things we’d like to learn about baseball. I’d like to expand on a few of the ideas, and maybe we can flesh out some thoughts to pursue, or at least plan to pursue them once we figure out how. The most intriguing ones, to me, were based on the concepts of performance being affected by teammates.

There were three suggestions that got at this kind of relationship, in different ways.

1. Catcher defense/pitch sequencing
2. Defense’s impact on developing a pitcher
3. Line-up synergy

In all three of these concepts, the idea is that one player is significantly impacted by the presence of another player. In general, statistical analysis doesn’t really account for any scenarios like that at the moment. We kind of throw our hands up in the air when it comes to catcher defense, and we create player projections in a context neutral environment and then add minor adjustments for things like park effects, but leave it at that.

From a practical standpoint, that’s okay for now. We don’t have any evidence that we should be doing anything differently, and you can’t just make up an adjustment for something that may or may not exist. But it’s not much of a stretch to think that there may be some kind of effects here that we’re missing.

The catcher/pitcher stuff is obviously ripe ground for study. With the accumulation of Pitch F/x data, we’re starting to get to the point where we can get some legitimate sample sizes, and start comparing what one catcher calls to others. We can look at trends of pitch usage and look for sequences of pitches that may be more effective than others. I think there’s a lot to be discovered in this area, and I’d expect it to be one of the major areas of study for statistical analysts in the next few years.

The interaction between a pitcher and his defense may be a little harder to study, but is also worth doing. Vince Gennaro, a professor at Columbia and a consultant to a lot of MLB teams, has recently done some work on the secondary effects of defense, quantifying the change in pitching staff usage for teams that flash the leather. They shift a lot of innings from their bad relievers to their good relievers and their starters, allowing for the distribution of innings to be more heavily skewed towards the top end of their talent pool.

However, there’s even further to drill on this issue. Do pitchers change their pitch selection based on the quality of their defense? If they have a bunch of gold glovers behind them, do they pitch more to contact? Does the resulting drop in baserunners result in lower stress pitches, which allow them to work deeper into games? These are benefits to the pitcher that have not yet been explored, but should be.

Finally, there’s the line-up synergy suggestion. This one, I’m a little less sure of, but would still think its worth our time to dive into. The protection theory has been studied to death and mostly debunked, but it isn’t hard to come up with scenarios where the performance of one hitter does affect others. It’s well known that nearly everyone hits better with men on base than with the bases empty, so anyone hitting behind an OBP machine should get a boost in performance, simply from that effect.

But there’s also other possible synergies, I would think. Having a balanced line-up of LH and RH hitters should limit a manager’s ability to play the match-ups late in games, reducing the amount of times a hitter has to face a same-handed pitcher in higher leverage situations. This would be especially important for a guy like Curtis Granderson, who should almost certainly hit between two RHBs. You could also argue for putting a left-handed groundball hitter behind a high on-base guy in order to take advantage of the hole created when the first baseman has to hold a runner on. These may not be huge changes, but they may add up enough to be worth considering.

While baseball is the most individual of team sports, it is not solely a one-on-one match-up at all times. If we look hard enough, I’d bet we’ll find ways that teammates do, in fact, influence the performance of those around them.

Print This Post

Dave is a co-founder of and contributes to the Wall Street Journal.

14 Responses to “Following Up on Some Questions”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Xave says:

    “We need to synergize this line-up.” “Yeah, really shift some paradigms.”

    Vote -1 Vote +1

  2. Adam says:

    One thing that I wondered is: could you assign a value to every pitch in a game to come up with the overall value contributed by the pitcher’s physical performance in that game? I’m not talking about balls, strikes, count, situation, and outcome. I’m talking about velocity, horizontal/vertical movement, location, pitch type, and sequencing.

    On a game-by-game basis you would have an outcome-independent assessment of how a pitcher performed. For example, take the hypothetical case that you have two guys that threw identical games and in the bottom of the ninth they both grooved a flat 88mph fastball, down the middle, up in the zone. In one game the batter jacks a 3-run HR while the other equal batter whiffs. One ends up a hero and one a goat as a result of exactly the same physical performance.

    Vote -1 Vote +1

    • Mike Fast says:

      Yes, you can. Jeremy Greenhouse has published on this topic at Baseball Analysts. Colin Wyers has also done some work in that arena, although not published as far as I know.

      Vote -1 Vote +1

    • Samuel Lingle says:

      That seems like it’d be EXTREMELY tough to do accurately. How are you going to account for “setting up” hitters? Will you give a fastball increased value if thrown after a few breaking pitches to throw off a hitter’s timing?

      I suppose if we get a large enough sample size we should be able to calculate how pitch effectiveness changes based in these situations so we’d be able to use some sort of average to help in those situations, but there are so many variables it seems like this would be exceedingly complicated.

      Vote -1 Vote +1

    • B N says:

      The issue with that approach is that it uses nothing about the context or the strategy. I have seen very good pitchers throw an obvious strike up the middle because they know a guy is looking to take a ball. You can’t assign a value to every pitch because there are too many factors that influence the value of that ball trajectory. For example, throwing balls out of the zone to Vlad is very different than throwing them to a guy like Cust.

      Basically, given that there is strategy- these approaches would penalize what might essentially be a “bluff.” Even imagining one single batter against a single pitcher, assume that a batter has one clear hole in their swing where they hit poorly at. Any “pitch-alone” metric would always rate that highest. But you can’t just throw the same pitch to the same part of the plate over and over and expect it to work. That’s called batting practice. Since the sequencing within an at bat and across the whole game is important, I just don’t think that kind of metric would have good insight into pitcher success. Batter success is a bit different because they don’t get the pick the trajectory, they just get to pick if they should swing at it. So while pitch to pitch data seems great for dealing with batters, for pitchers I think it is only useful to expose glaring strong spots or weak spots but not all that useful for cross-pitcher comparison.

      Vote -1 Vote +1

    • B N says:

      One thing that might be interesting though would be a plate appearance Markov Chain analysis looking at the pitch types, batter actions, and outcomes though. Looking at that kind of data might give some idea of the risk/reward of a particular pitch in a particular circumstance. That would provide the proper negative weight to throwing the same pitch over and over, while giving some insight into if a pitcher was taking a good bet or a worse bet on a particular pitch (given their history and the batter’s history).

      Not sure of exactly the variables I’d want included in the Markov State, but I’m sure there are a bunch of valid and interesting choices.

      Vote -1 Vote +1

  3. Basil Ganglia says:

    “Do pitchers change their pitch selection based on the quality of their defense?”

    Maybe we should just look at Brian Bannsiters’s GB% with and without Jose Guillen in the OF?

    Vote -1 Vote +1

  4. Nate says:


    I missed yesterday’s open question thread and would like to add a late addition.

    Are we too quick to attribute a pitcher’s BABIP to Luck? With Pitch F/x data we can now ask if a pitcher has a tendency to put himself in a position where he could give up more hits than average. The basic premise of BABIP = Luck is that a pitcher has little if any control over what happens once the batter makes contact. I’d argue that a pitcher who grooves a fastabll when he has an 0-2 count on the hitter has made the outcome entirely dependent on Luck; while a pitcher who throws his fastball on the outside corner has added a measure of skill to the equation.

    I realize this might be a labor intensive project, but can we refine the data to show us which pitchers are making good pitches and are getting unlucky vs. the guys who are just getting lucky?

    Vote -1 Vote +1

    • Samuel Lingle says:

      I’d really like to see a lot more analysis of this.

      Teams can set up defense to expect a groundball to one side of the field and a pitcher can pitch to produce said groundball. I’ve read somewhere that Joe Nathan’s BABIPs are usually low because he induces an inordinate amount of popups to the infield. It seems like there are a lot of things here that could have a noticeable BABIP effect that we ignore a lot of the time.

      Of course, how much is it the pitcher’s skill if he’s pitching to favor his defensive alignment versus the defense’s skill or even the coach’s? Seems like it’d be hard to separate them from each other in this case.

      Vote -1 Vote +1

      • TCQ says:

        It’d be great if we could get something in the areas mentioned, but the Nathan example doesn’t really work. He gets a lot of pop-ups – that’s part of his skill set, all the time. The idea that certain pitchers are able to “change” their skill-set based on the defensive setup is plausible, but entirely different, I think.

        Vote -1 Vote +1

      • Samuel Lingle says:

        I didn’t really mean those as two examples of the same thing, but two examples of things that could differentiate different pitchers’ BABIPs. I’m sure there are plenty of other factors that go into it that they control.

        Maybe overall these effects are negligible, but at this point it seems like we don’t know enough to prove it.

        Vote -1 Vote +1

  5. Colin says:

    I had been thinking about the interaction between pitchers and defense. If you hypothesize that a pitcher bases their pitches on their defense, a large fundamental gap that seems to come up in measuring this is the gap between what a pitcher thinks their defense is capable of of and what their defenses true talent is.
    I’m willing to bet that a pitcher’s faith in their defenders in based much more on the subjective standards that gold glove voting are based on then UZR (or comparable stats.) It seems like you would almost need blinded pitchers to sort this out.

    Vote -1 Vote +1

  6. Circlechange11 says:

    I’d like to see some pitcher data regarding what % of their IP see them allowing a ‘crooked number’. In other words, a P’s “ability” to limit damage if such a thing exist.

    We all watch games where a pitcher seems to be cruising and before ya know it he is getting knocked around and out of the game. That’s one thing I’ve wondered about pitchers that over/under perform their FIP.

    Vote -1 Vote +1

  7. Brendan says:

    Wouldn’t it be a bad idea to put a GB hitter behind an OBP guy, assuming all else is equal? Granted, there would be some extra room in the 4-hole, but doesn’t that set you up for a lot of double plays?

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current ye@r *