THE BOOK--Playing The Percentages In Baseball

I’ll run the numbers on Webb and Zito (maybe a few others) either later tonight or tomorrow.  Sounds l interesting way to look at ground ball pitchers. 

Yeah… and scratching the surface is pretty much all I did.  Guess we really need to find that billionaire.

Yes, that is one of around 10,000 interesting questions.  I asked that question about sinkerball pitchers in general on BTF a while ago.  I know a lot about pitching, but I have no idea whether there is something about the way a sinkerballer throws the ball that enables it to drop more than non-sinkerball pitchers or whether a sinker is merely a two-seam fastball thrown down in the zone (IOW, sinkerballers merely throw more pitches down in the zone).  Most of the response were that they thought a sinker was merely a low two-seam fastball.  There is only three ways to get a ball to “sink”.  One is less speed, another is less rotation, and the third is by the axis of rotation (a ball with overspin, like an ovrehand curve, sinks the most).  Maybe sinkerballers have less natural rotation (like a splitter).  There is actually a third way, which is orientation of the seams which is why a sinker is usually a two-seamer.  I think that a two-seamer drops more than a 4-seamer given the same speed.

Anyway this is the same question that Tango poses.  If a sinker is merely a low fastball, it is odd that commentators (and coaches, pitchers, etc.) always say that a sinker that does not sink is a very hittable pitch.  If a sinker is merely a low fastball, then a sinker that does not sink is merely a high or medium height fastball and should be no more hittable than any other non-low fastball.  It sure does seem that when Carlos Slva throws a 94 mph sinker up in the zone that it gets hammered.  Of course there are many things that go into whether a ball tends to get hammered or not other than location and speed - deception by the pitcher, movement of the ball (from spin), not to mention how predictable a pitch is.  For example, Silve throws almost exclusively sinkers (fastballs) and does not change speed much.  In fact, most sinkerballers do not change speed much.  So the batter is almost always geared up for the same speed.  Given that, when the ball is up, it probably should be hammered.  I actually never thought about that before now.  Maybe if a sinkerballs varied his pitches more (curve, slider, etc.), when he throws a sinker that does not sink, it would not get hammered as often.

Oh, and I am concerned that there might be bias in the recording of the data.  I hope not.  Could it be that when a ground ball is produced the scorer tends to record the pitch a little lower in the zone than if a fly ball was produced, even if they were thrown to exactly the same location?

Could they be doing the same thing with “known sinkerbalers”?  Could sinkerballers actually have the illusion that their pitches “sink” more than usual and thus the scorers again input lower then actual location?  If a pitch does in fact “sink” could the scorer be nputting a too-low location and vice versa for a 4 seam fastball that gives the illusion of rising.

I think these are very real issues that need to be considered otherwise if these biases exist, there are going to be some erroneous conclusions drawn from studies utilizing this data.  I would much rather see cameras and computers recording pitch data than individuals using their judgement (I assume that is how it is done by BIS).  There is too much potential for significant bias on many fronts especially the outcome of a pitch affecting how the scorer perceives the location.

I just RTFA.  Fascinating stuff and I agree that this is just scratching the surface and an ever so slight scratch at that.

I am shocked that there is not a higher correlation between upper/lower third % and G/F ratios.  I would have thought that a pitcher’s G/F ratio was almost completely a function of where on the vertical axis the ball tends to be thrown.  Apparently not.

I am also very shocked that there is not a significant y-t-y correlation in % of pitches thrown to middle of the plate.  Shouldn’t pitchers with less control be much more prone to throwing pitches in the middle of the plate?  Shouldn’t pitchers, regardless of their control, have a deliberate tendency to be nibblers or not?  I mean, if I have great stuff like 1 97 mph fastball, aren’t I going to have a tendency to throw the ball near the middle of the plate and vice versa if I don’t have great stuff (I want to throw to the corners)?

I also was surprised by low y-t-y correlation for middle of plate.  That may be because it’s only one of 25 cells, so the range from ‘good’ to ‘bad’ is something like 8% to 12%, which is hard to find among the noise in y-t-y data (similar to line drive %).  Given several years of data, we might find differences among pitchers.

A variation on Tango’s idea is to see if pitchers do better when pitching to their strength, compared to other locations.  Do bottom-third pitchers pitch better there but get hammered up?  Do outside-third pitchers do well outside, but get lit up inside?  Or do they reach a balance where they get hit about equally well in all locations, which seems like what should happen.

For what it’s worth.

NPB 2006 season data

BBT breakdown for RH batters, excluding pitchers

Pitches within the strikezone

8 - middle high
5 - middle
2 - middle low

Bats    BIP    L%      F%    G%    Loc    PitchType


R    1251    .224    .421    .355    8    Fastball


R     799    .213    .432    .355    8    Other


R    1289    .265    .348    .386    5    Fastball


R    1045    .303    .332    .365    5    Other


R    1025    .231    .316    .453    2    Fastball


R    1209    .245    .311    .444    2    Other

Fastballs include fastballs and tailers (shuuto).

(See Website link for help on notation and more - posts by Sergei)

Sergei, where is the data/article/etc. on the website and what raw data are you using?

Also, what is a tailer (shuuto)?

MGL,

> Sergei, where is the data/article/etc. on the website ...

Oops, the link actually appeared under my name, sorry for confusion.
It’s an older Batter’s Box tread named Slumping Carlos: Jackal and Hyde? (again, posts by Sergei).

Most importantly,

[pitch locations] are coded 1-9 as follows (from batter's point of view):





7 8 9


4 5 6


1 2 3





For example, 7 - high inside corner for right-handed hitter.

I’ll add BBT breakdown for other parts of the zone later today.

> ... what raw data are you using?

Pasting from a JapaneseBaseball.com discussion:

“BBTs, pitch-by-pitch aggregate data, is not openly available online.  You’ll need to gather game data, at-bat by at-bat, on a daily basis throughout the season and process the bits into database format, just like I did.”

You can see some pitch graphs there as well.

> Also, what is a tailer (shuuto)?

Tailing fastballs.

See this great clip
http://japanesebaseball.com/video/shoot.wmv
found in the Gyroball/Shuuto/"The Ghost Pitch” tread.

NPB 2006, right-handed batters, pitches in the zone.
Batted ball breakdown by location and pitch type.

Bats    BIP    L%    F%    G%    Loc    PitchType


R    1502    .202    .365    .433    9    Fastball


R    1251    .224    .421    .355    8    Fastball


R    1509    .176    .447    .377    7    Fastball


R    1722    .234    .287    .479    6    Fastball


R    1289    .265    .348    .386    5    Fastball


R    1642    .214    .346    .440    4    Fastball


R    1207    .196    .239    .566    3    Fastball


R    1025    .231    .316    .453    2    Fastball


R     895    .212    .293    .495    1    Fastball





Bats    BIP    L%    F%    G%    Loc    PitchType


R    1206    .180    .343    .477    9    Other


R     799    .213    .432    .355    8    Other


R     344    .186    .398    .416    7    Other


R    2358    .213    .304    .483    6    Other


R    1045    .303    .332    .365    5    Other


R     529    .197    .391    .412    4    Other


R    2424    .171    .276    .553    3    Other


R    1209    .245    .311    .444    2    Other


R     413    .208    .351    .441    1    Other

Note: unfortunately BIP include bunts, that didn’t become sacrifice hits, thus all GB percentages should be somewhat lower. Another nitpick is that the Japanese (DataStadium) score many home runs as line drives.

If someone wants to help come up with a list of “sinkerballers”, I can run the numbers on high/middle/low pitch outcomes later this week.

Westbrook, Lowe, Webb, Wang… who else?

I think DSG had a list compiled (I think from ESPN.com).  If he’s around, maybe he can help you out.

DSG came through and the data is on its way. I’ll post an update by next week.

Didn’t get around to looking at groundballs by location until today:

Pitch Location & Groundballs

Seems pretty clear at least in the Webb & Zito case that there is such a thing as a “true groundball pitcher”. 

Just thinking aloud here, but I bet there’s some sort of self selecting process with pitch location and the pitches you throw.  I’d suspect that where you throw your pitches is more indicative of what types of pitches you throw and your “pitching style” than anything else.

Wow, great job David, a real eye opener!

David continues scratching:

http://baseballanalysts.com/archives/2006/12/expanding_the_s.php

Bumping for you ball_trackers who may have missed it the first time.