THE BOOK--Playing The Percentages In Baseball

Link? Sauce?

As a Phillies fan, I’d welcome anyone to put Utley on to face Howard and then the lefty slaying Werth.  Much of the problem the Phillies have had is that their homeruns are mainly going as solo shots.  I believe all but Utley’s 3 run dinger were solo bombs.  Give them more baserunners by walking Utley opens up the opportunity for more multi-run homers.

Tango,

I think this really echos how important MGL’s pitch selection piece was the other day.

Buster always gets me flustered.

One of the new things I did this year was to cross-reference pitch types with home runs, using PitchFx pitch classifications where available, and then selectively viewing video to fill in the gaps where I wanted to have 100% data.

Since June 9th, Utley has hit 19 regular season and 6 post-season home runs, for a total in that time period of 25.

Of those 25, 21 have been hit off fastballs (20 off four-seamers, 1 off a 2-seamer), 3 off sliders, and only 1 off a curveball (that being his homer off C.C. Sabathia in WS Game #4.)

Put another way, of all the curve balls thrown to Utley in the last 5 months, he has only hit one over the fence.

To me, increasing his daily intake of curveballs seems like a valid approach… it’s no guarantee that he won’t hit any out, or won’t hit any safely for hits, but then I don’t think Buster Olney was suggesting that a steady diet of breaking balls would definitively wipe Utley out, either…

Oh, and I should probably tell what Utley hit out before June 8th:

FC:  1
FF:  5
CH:  3
SL:  3

So, he hasn’t hit a changeup out of the park since June 1st, either.  Maybe that should be part of the diet, too…

Buster is certainly a tool, but this is the kind of thinking that is pervasive in the industry as well, among insiders - that teams now have to completely change the way they pitch Utley because he has been “hot.” If teams/pitchers actually did that, batters would kill them.  Basically Utley would go from being pitched relatively optimally according to his strengths and weakness garnered from years of scouting (which is constantly updated - I am not suggesting that the optimal pitching approach to batters never changes - it does) to being pitched completely sub-optimally based on 10 games of being “hot” to the point where that approach would be so predictable to him that he would take tremendous advantage of it.

Again, can you imagine if all of a sudden they started throwing him 70% curve balls when they used to throw him 40% (or whatever it was) curve balls which was optimal in the first place?  He would kill that approach.  And if that new approach were correct than it would have been more or less correct a month or 5 months ago.

It amazes me and strangely saddens me that Andy Pettitte will almost certainly take the hill and not have the PitchFx data that Greg just provided.

If major league managers were civil engineers, I would never drive over a bridge.

Funny. I don’t have time to dig up all the examples. But Buster’s solution* to netralizing ANY “hot” hitter is to “bust him inside with fastballs”. Every time!

*Actually, he usually quotes some unnamed scout or front-office type

While I totally agree that a pitch mix that included 70% curveballs would certainly not be optimal, I also doubt that pitchers are using an optimal mix in the first place.

But yeah...if you are going to change your approach to Utley, don’t change it based on his last 30 plate appearances. Change it based on data that goes back hundreds of plate appearances.

Wow, Greg, there are many other things that you would have to look at before deciding that he needs more curve balls. Many, many other things.  The data you provide doesn’t even begin to scratch the surface.  For example:

What percentage of HR’s for all LHB are hit off the various pitches?

What was the handedness of the pitchers for those HR’s?

What was the location of those fastballs?

Do all we care about is HR’s when determining how to pitch to batters?  Of course not!  What if he only hits HR’s off fastballs but never anything else and never walks on them ?  What if he always swings and misses on fastballs but occasionally hits a HR (I am exaggerating of course)?

What is the average value of each pitch thrown to him?  Has that changed lately and is there any reason to think that is not a fluke?

What about in prior years?  Is there any reason to think that his true talent against a certain pitching approach has changes since June 8th?

Where did you even get that date from?  Did you cherry pick it?

While the pitch f/x data is immensely valuable for deciding how to approach batters AND, as I said, the optimal approach to batters can change over time, it is going to take a lot more than just looking at on what pitches a batter hits his HR’s before and after a certain date to make those decisions, and I highly doubt that you would EVER radically change an approach to a batter at any point in time - certainly not one that has been in the league for more than 5 years and whose overall true talent has not appeared to change much in those 5+ years.

“Steady diet of curve balls?” No chance that is correct!

I am with BrianK, I doubt that Utley, or any hitter, is being pitched optimally.

It would be rather unlikely for that to be the case, because there are two people to consider, not just one.  What is an optimal breakdown of pitches to go after Utley with (say 60% fastballs, 20% curves, 20% change, to make something up) is almost certainly not going to be the optimal breakdown for the pitcher who is throwing to him (say 70% fastballs, 20% sliders, 5% changes, 5% curves), and when those two don’t match, the ideal answer will necessarily be somewhere in between the two.

MGL #10

Sure, I know you need to look at more than just what I cited.  I’m just tossing a data point into the discussion, I’m not trying to claim that this data point settles it.

And I don’t think all curveballs would be the right approach, either.  Obviously if you went all the way to that extreme, you’d lose the benefits of deception and disruption of timing, eye angle and so on that a mix of pitches provides. 

I doubt Olney meant it literally, either.  A “steady diet of ___” is a colloquialism that means “more than usual”, maybe a lot more than usual.

And I did “cherry-pick” the June 8th date, as you surmised, but subsequently, overcome with guilt, I filled in the blanks for the rest of the season.  However, my interval was 2/3 of a season, and runs up to the current date, so it wasn’t a particularly egregious example of cherry-picking, I think

I have no doubt that when Barry Bonds started becoming BARRY BONDS that pitchers decided that they needed to throw to him differently (more “optimally").  And the result is that he started walking even more.

The “optimal” approach is basically not to pitch the guy anywhere close to the center of the plate, such that you will end up walking him tons.  You reduce his HR (and in fact all his contact plays), and increase his walks.

If by “optimal” it means “less HR”, then yeah, not throwing fastballs down the middle will help.  But, “optimal” means “as few runs as possible”.

You can optimize a little bit, but not that much.  Any wholesale change will undoubtedly lead to a suboptimal approach.

Sure, Tango, but remember that over the next two days, the guy hitting after Utley (Howard) is not going to be facing two average pitchers, he is going to be (predominantly) facing two significantly better than average left-handed pitchers, so the “optimal” approach to attacking Utley in this particular situation is not going to be the same as it was all year. 

If you think (with data to back it up) that you’re more likely to get out Howard than is normally the case, then you’re happy to shift Utley’s production distribution away from homers and towards walks.

Forgive me if I missed this, but what percentage of each pitch did Utley see this year? I know it won’t account for all of the difference in the homeruns by pitch type, but it must account for some of it, right?

I guess what I’m asking is this: how different is his HR on different types of pitches? (I’m assuming the difference might be less than his HR by pitch type for one season suggests...)

Am I out to lunch?

Per Fangraphs pitch type values:
http://www.fangraphs.com/statss.aspx?playerid=1679&position=2B

Fastball/Cutter 63%, +.015 runs/pitch
Breaking Ball 24%, +.009 runs/pitch
Change/Split 13%, +.013 runs/pitch

Those breakdowns obviously vary with LH and RH pitchers and of course the repertoire of the pitchers. I am glad to see all the pitch values about the same, as they should be, more or less, according to the “minimax” tenet of game theory.

I doubt those percentages are going to change much tonight, other than because of the way Pettitte (and the relievers) pitch normally.  Why should they?  My basic point is that if pitchers/managers change their pitch selection to hitters based on their hotness and coldness and those around them in the lineup, they are making a big mistake. 

My other point is that even though managers obviously do put faith in hotness and coldness and let it dictate pitching “approaches” (like IBB’ing A-Rod with the bases empty), I sincerely doubt it is going to change the way they pitch Utley pitch-selection-wise, very much at all.  They are NOT going to throw him a steady diet of curve balls - at least I don’t think so.  And I realize that even Olney does not mean ALL curve balls.

Mike F., do you mind showing your work?  I got a much smaller value number for the CH and SF’s, so I want to make sure I am calculating it right.  It looks like you may have used a positive value for Utley’s hitting against SF’s, instead of the actual negative value for 2009…

I took a quick look at Pettitte, and it looks like his highest value pitch is a curve ball, so I want to be doubly sure of the calculation before I try to make anything out of it…

Greg, I used Utley’s career values. 

There is some variation year to year, but I didn’t see a particular trend where Utley was getting dramatically better or worse over time against breaking balls, so I think the career values are a better approach than just looking at the small sample of 2009.

OK, Mike, I agree with that, there is definitely a lot of variation in the single year numbers.  It just happened that a couple of the numbers I checked first from 2009 seemed to match yours, so I ran my numbers for 2009.

However, that leads me to my next question, which is for MGL:

How do you figure that 0.015 is about the same as 0.009?  Is this a variation argument?  How much uncertainty do you figure there is in the numbers for Utley’s career (well, the portion of it that overlaps our pitch type data)? 

Nominally, they are quite different…

In any event, I look forward to seeing how many curves Pettitte throws Utley tonight!

I really didn’t think about it other than .009 is pretty close to .015, isn’t it?  Honestly, I don’t know either the standard error for those numbers for that sample size and I don’t know the SD in talent for those numbers (the typical true talent among players). Without knowing those two things, you can’t really say that one number is “close” to another. I was just speaking off the cuff.

I should give some context on those runs/pitch numbers.

For hitters, if you are at +.020 runs/pitch against a particular pitch type, you will be among the league leaders.  If you are at -.010 to -.015, you will be among the league worst against that pitch type.

The spread is similar for pitchers but differs a bit for specific pitch types and for relievers vs. starters.

Ah, I see MGL/#24 and I were thinking along the same lines. 

I don’t know the standard error for these numbers.  You could estimate the number of pitches in each sample by saying that Utley’s seen 5 yrs * 600 PA * 4 pitch/PA = 12,000 pitches in his career, then take the percentage for each type of pitch.

I didn’t exactly give the SD in talent in Post #22, but I’d guess it’s somewhere just south of .010 runs/pitch.

Normally on an analysis I’d roll my own numbers, but in this case it was easier to take them from Fangraphs for a quick and dirty estimate.

If I were pitching tonight for the Yankees, if anything I would throw Utley MORE fastballs than I typically would…

That helps, Mike, thanks.

So much for a “steady diet of breaking balls.”

So far at least…

Too bad I can’t reply to my own post.  Tonight’s game had a shining example of why it’s dumb to avoid Utley to pitch to Howard.  (Un)fortunately it didn’t affect the outcome of the game.

I counted 19 total pitches to Utley: 10 fastball/cutter and 9 curve/slider.

That’s 47% breaking balls, about double the season rate of 24% Mike cited above.

FWIW…