THE BOOK--Playing The Percentages In Baseball

Is there a link to the study?

This would seem to combine very neatly with the Pitch framing study from Mike Fast that also came out recently, and give a nearly complete reading on catcher defense. Assuming the results hold, I suppose plate blocking is left to figure (a rare but high run impact play) out, but maybe the player rating systems can start feeding this stuff in for catchers soon.

I’d be curious to know if pitchers are naturally aware of the catchers abilities to block pitches -do the worst blockers get different pitches thrown to them? I know catchers “call” pitches, but pitchers shake them off a lot, so the effects would be entangled. At a glance, it appears the the best blockers have a higher cPP rate than the worst.

Bojan’s article is here:

http://www.hardballtimes.com/main/article/another-one-bites-the-dust/

Aweb/1, what we’re still missing, beyond fielding throws at home plate and/or blocking the plate, is a measurement of pitch calling performance.

We can get a vague idea of that by looking at catchers who perform well in a multi-year catcher ERA (like Sean Smith developed) as compared to how they perform in getting extra strike calls. So we can guess that guys like Jason Kendall who rank well in catcher ERA but poorly in pitch framing must be good pitch callers.  But we would very much like to have a more direct measure of pitch calling performance.

awesome stuff.  wow. 

it makes me wonder about studying the same thing except from a pitcher standpoint.  ie the pitchers who give up / save runs on the amount of wild pitches and passed balls they are at fault for.

Wow...and I thought the sun position stuff a few days ago was a creative data set and graph. This is really, really impressive.

Mike Fast/3
Pitch calling will be especially tricky, because:
1. some managers/pitching coaches (?) call the pitches from the dugout. This happens with young guys especially. Who gets credit? Like positioning defenders, it’ll be hard to extract player from coach.
2. Pitchers shake off catchers constantly, and there is no way to know what pitch would have been thrown. I suppose part of the skill could be seen as getting a pitcher to throw the pitches you want, but it will be very hard to pull apart.

I think pitch calling, properly accounting for quality of pitcher and opponent, might be best reversed engineered from everything else that is starting to come together, as you seem to be saying. Catcher ERA (or RA to be sure to not exclude catcher errors) would include pitch framing, pitch blocking, running game control, blocking the plate and pitch calling (and probably other things I forget). Once you have everything else figured out (it’ll be a while, but it seems at least plausible), maybe pitch calling falls out the other end as whatever is left?

Great stuff…

This is cool. Not surprisingly the data supports what we generally know/knew or makes the most logical sense.

Balls in the middle are easiest to block.

Balls in the dirt glove side are easier to block than non-glove side.

The farther in front of the plate it bounces, the tougher it is to block.

I would guess that you could take the same idea and apply it to infielders on grounders and the shading would be similar. Just interesting.

The data goes a step further and quantifies the information.

I’m amazed that MLB catcher block 70% of th eballs non-glove side that bounce far in front of the plate. I cannot begin to describe how difficult that is.

I’m going to point out that stat to every dad at a baseball game that keeps riding their youth catcher because they only block 3-5 out of 10 of those type of pitches. Damn, yell at the pitcher for missing his target by THAT much.

Anyway, this is beautiful stuff.

Tom, thanks for the kind words.

I’d like to thank both you and Mike for the help and encouragement to do the research.

aweb/1

I agree, looking at the issue from the pitcher’s perspective could answer some interesting questions. One is, as you point out, which pitchers are in general tough to catch? And second, as I stated at the end of the article, do pitchers recognize certain catchers as good blocking battery mates and trust them more than others? In other words, is a pitcher more likely to throw a tough low breaking ball on a 0-2 count and a runner on third in a high leverage situation to Quintero than he is to Johnson?

"The data goes a step further and quantifies the information. “

This is really what we do.  Everything is fairly obvious in terms of “direction”.  What saberists do is give you “magnitude”.

There’s no surprise that a catcher will block more on the glove-side than the nonglove-side.  The question is how much more.  And, thanks to Bojan, now we see it all.

This is true of everything.  We know that clutch hitting exists, we know that pitchers control hits on balls in play, we know that protection exists, we know that pitchers can have on and off days.  We know this is true because players are human, and humans respond to their environment in non-random ways.  The question is the degree to which it exists.  If you can’t find SOMETHING, then you’re not looking hard enough.  And then once you figure out the degree, that’ll tell you if it’s actionable.

Clutch hitting, for example, is barely actionable.  It’s a tie-breaker kind of ability.  At MOST, if you have a great clutch hitter and you have a poor clutch hitter, but the poor clutch hitter has the platoon advantage, then it’s a tie-breaker, at best.  Good luck finding out who has the clutch skill though (because the numbers won’t help you).

Interesting to me that Yadier is great, and Jose and Bengie suck.

This is really what we do.  Everything is fairly obvious in terms of “direction”.  What saberists do is give you “magnitude”.

Yeah, I was kinda stating the obvious there. heh heh.

Did I tell you about my uncanny ability to tell whether Carpenter is going to give up runs over the last 2 innings or not?

This is really what we do.  Everything is fairly obvious in terms of “direction”.  What saberists do is give you “magnitude”.

Maybe I’m interpreting this statement differently than Tango intended, but, to me, the best quantitative studies shift the way we qualitatively think about things. There are often unintended consequences of research that are way cooler than finding the magnitude of what’s being looked at.

I’m not sure I can think research right now that has that sort of paradigm shift.  Can you give an example?

By the way, my help to Bojan amounted to little more than making a reference to Star Trek about “thinking three-dimensionally”.  As great as his article is here, I’m more shocked that he never saw Wrath of Khan.

Just amazing to me that I get to read this kind of stuff for free.  Reading work of this quality is, I know, the closest I will ever get to the inside walls of an MLB front office, and I am grateful to those - including this website’s hosts - who take me there.

Very cool.

Tom/14

It’s a dark spot on my curriculum and my right to wear a supernerd’s cape was revoked based on that sad fact.

Bojan: how far back in time do your sci-fi bonafides go?  And is there some European-culture difference?

I’d guess that overall, Europe has less of a sci-fi attachment than the States, although this might just be my personal experience. Although, FWIW, one of my best friends always used to say to his little daughter “This Star Wars figures are daddy’s toys and you know what’s gonna happen if you touch them”. Or something like that.

I never watched much TV, although I remember loosely following Battlestar Galactica when I was a kid. I know, I know, I’ll go kneel in the corner now.

Reminds me of Patrick Stewart / Ricky Gervais:

http://www.youtube.com/watch?v=Fg_cwI1Xj4M

You can watch the whole thing as it’s funny, but the relevant part is near the end.

Phew! I have an excuse, I’m married!

Actually, I even got my wife to draw the catcher for me. Which is not a bad thing, as she draws for a living, while I get constantly asked by our nieces which animal it is that I just drew.

As an example:

http://holzfeder.com/gallery/THT/3d-plane-bojan.jpg
http://holzfeder.com/gallery/THT/3d-plane-vir.jpg

Great work Bojan. Would definitely like to see the same thing for pitchers, and then you would have to adjust the catchers for the pitchers they catch and vice versa using an iterative method or one of those computer programs that is able to “automatically” tease out the relative contribution of both agents in something like this (or stolen bases and infield GDP)…

It’s the exact same process as WOWY.  You don’t have to do much iteration.  At most two iterations is enough.

This is no different than strength of schedule.

By the way, next time I need an artist, I’m going to put your wife on my speed dial.

Given that Bojan controlled for pitch speed and pitch type, I wouldn’t think that identity of the pitcher would be a significant biasing factor in his results.

What he couldn’t have controlled for is the intended and actual location. 

If you have a pitcher who is just wild, and you have another who is not, but both throw the same repertoire, the guy who is wild is going to get more uncaught pitches.

I don’t know how significant it is, but it’s easy enough to control for.

I wonder what % of the pitches that bounced on the corners of R1 were breaking pitches. My guess is A LOT, as in the large majority, as in perhaps 70+%.

So, not only are catchers going to [1] move the greatest distance, [2] get an “in between hop”, but [3] they’re going to have to deal with serious spin that doesn’t always behave the same way.

I love the red corners in R12. “I wouldn’t dig in if I were you. I don’t know where it’s goin’. Swear to God.” You know the movie.

"Given that Bojan controlled for pitch speed and pitch type, I wouldn’t think that identity of the pitcher would be a significant biasing factor in his results.”

Right, good point.

“What he couldn’t have controlled for is the intended and actual location.”

Also good point.  Surely a pitch in the dirt or to some other extreme location from Maddux, where the location is expected, is going to be a lot easier to catch than one from A.J., which is typically not expected and could have been intended to be thrown to a spot 3 feet away…

mgl/21
mike/24
tango/25

Thanks, MGL.

You guys are all correct - controlling for the _unexpected_ location is the biggest remaining problem of this study, although most other personal effects of the pitchers should be covered in the location and speed. So, perhaps home/road splits, too?

Another thing that occurred to me is the park effects - possibly together with the game time. You know these creepy shadows that make it hard to see the ball coming in.

I will tackle pitchers next and then do a WOWY.

I wouldn’t worry too much about home/road splits although it would be nice to see the overall split.  Individual H/R splits are not necessary.  Park effects might be nice, though, as you say.  You want to do that the same way we do any park effects.  Home and road teams combined, home and away. Then a correlation (ICC, year to year, etc.) to see how much of the park effects are “real” and how much is noise.  Do you do dishes too?

Well it’s either baseball or dishes, so I’m glad I’m finding a way to prolong the baseball season, if only the stats aspect of it. My wife may or may not share my enthusiasm about it.

When I come to doing WOWY and park effects, I might ask for some more specific advice, as you guys have ton of experience with it.

While you are at it, you may as well throw in day/night.

You guys are all correct - controlling for the _unexpected_ location is the biggest remaining problem of this study, although most other personal effects of the pitchers should be covered in the location and speed. So, perhaps home/road splits, too?

I think that the results for knuckleballs would indicate that the unexpected factor is by far the most important factor.  Knuckleballs shouldn’t be any harder to catch than any other pitches for any other reason.  In fact they should be easier to catch, since they are low speed and high speed was identified as a factor that decreased catchability.

Knuckleballs have unexpected movement, which seems like it might be a lot harder to deal with than a pitch with expected movement but in an unexpected location.  For non-knuckleball pitches, the latter is going to happen a lot more often than the former.

The frequency with which something happens is important, too, not just how much it increases the difficulty factor.  If a pitcher and catcher get crossed up only once or twice a season, it doesn’t matter too much if that increases the likelihood of a wild pitch by 2x or by 100x.

I don’t know how often pitchers and catchers actually get crossed up, but I know that pitchers miss their intended locations all the time, many times a game.

If a pitcher and catcher get crossed up only once or twice a season, it doesn’t matter too much if that increases the likelihood of a wild pitch by 2x or by 100x.

Mike - All the rates for passed balls are very small.  There are only 12 to 13 passed balls separating Quintero at number 1 from Torrealba at number 15.  So you are right that 1 or 2 a season won’t make much of a difference, but 15 a season would make a very big difference.  That’s only 1 pitch every 8 games which seems within the realm of possibility of variation between pitching staffs to me.

I agree that identity of the pitcher is a necessity to look at, not to mention pretty easy to handle.  Whether the results are going to change by more than 1 or 2 runs, well, let’s see! 

(Presumably, if you have a pitching staff, and one guy gets at most 15% of the pitches, then chances are most pitching staffs are going to be a wash.)

I’m not sure I can think research right now that has that sort of paradigm shift.  Can you give an example?

OT, but what about the quarterback’s role in taking sacks and throwing interceptions?

I don’t follow football nearly enough.  But what is so special about the QB, sacks, and throwing interceptions that we didn’t know qualitatively before?

My argument, in case it wasn’t clear, is that the fundamental way to play hasn’t changed (in baseball anyway).  That qualitatively, we know how things work.  Then it’s just the quantitative aspect that moves the breakeven point.

Take for example the SB.  We have a decent idea of when to steal and when not.  In order to get a better handle, we need data, so that shifts the breakeven point to the right spot.

I suppose in football, whether to have one RB or two would be a paradigm shift.  What role the TE should have maybe.

In basketball, zone v man-to-man I think would be a paradigm shift.

Hockey?  They went through one decades back, going from up-and-down to a criss-crossing pattern.

Those are real team sports though.

Ok, I have one in baseball: the shift.  That I think is a qualitative one.  Relievers each given an “inning” as opposed to a “leverage”.  That might be another qualitative one.