THE BOOK--Playing The Percentages In Baseball

With all hands on deck, with enough people in the bullpens, we should really probably see the pitchers pinch hit for every time their spot comes up. That’s going to be too much to ask for, but maybe they’ll actually PH their second time up? We can hope.

I don’t have my copy of The Book handy.  How much do you knock off of Carpenter’s projection because he’s starting on 3 days rest?

A 3/2/1 of his FIPs gives you a projection of 3.22 FIP.

Is Washington trying to run his way out of this game?

If they keep getting all those hits, it won’t matter. But Kinsler’s pickoff cost at least 1 run.

Isn’t that the 2nd time this series Kinsler has erased himself after getting on leading off the game?

At least one of those times, quite possibly both, Washington has gotten Kinsler erased. It looks like he wasn’t comfortable with the infield here, though.

Okay - the Kinsler pickoff matters. In large part because of a quasi-intentional walk with 2 outs to Pujols.

Pinch hitting for Harrison in the 2nd here, probably questionable, close, not going to be too hard on Washington for going with the conventional wisdom leaving him in in the 2nd. (BTW, is there any count where a bunting pitcher gets a good enough count bonus to switch to swinging away?) On the other hand, Carpenter looks like he has NO command at all, and I’m willing to say that TLR should definitely take him out as soon as his spot comes up in the order, *possibly* sooner if that’s not here in the 2nd.

I thought Harrison looked bad enough in the first that I would have definitely pinch-hit for him in the 2nd.

Also, that walk to Pujols was ridiculous. I wonder if it affected Harrison, but I suppose he hadn’t really thrown enough pitches to get a read on his command.

Furcal’s swing is too big. Little guys should have small swings. I mean, if that Pedroia guy wouldn’t swing so hard, he might be able to be a starter some day. Sorry. Forgot to mute before joining the chat.

Argh!!!

As you can see, Carpenter is just not “on” tonight!

Anyway, there is nothing more pathetic than watching someone bunt at a 2-1 pitch a foot out of the strike zone. One of the reasons you DON’T bunt with a 2-1 count, besides turning the batter into a much better hitter, is that you tend to bunt at balls that you would take if you were not bunting.

No mention by McCarver or Buck about bunting at a 2-1 pitch a foot up and away…

I haven’t mentioned this in any of the other threads, but did anyone notice that before last night’s game, almost everyone in the media was calling Washington a genius?

Last night he was a bumbling idiot.

Tonight, we have to wait and see who wins before we know who the media is going to say out-managed whom…

Shouldn’t it be the little guy who HAS to swing hard?

No one in the mainstream media’s equipped to criticize a manager absent the results of his team.  Oh, they may try, to the tune of “they won in spite of him,” yet that’s as far as it goes.

That was one of the worst IBB’s you will ever see.  Of course I’ve said that many times.

Do you think that Ron Washington even knows that when you IBB the bases loaded, that if the next guy walks, it’s a run?

I don’t care how good a people person he is, the guy is a stone cold moron!

I wish that I were part of the mainstream media so that I could tell it like it really is…

Argh! again

You have to PH for Carpenter there.

Carpenter batting in the bottom of the 6th.  Game 7 of the WS. 

This can’t be correct.

I agree - especially with Carpenter batting leadoff. Maybe TLR is saving the bullpen!

... or maybe not. Carpenter gets pulled with nobody out. But I give TLR credit for treating the failure to pinch hit as a sunk cost now.

Why would you pinch hit for him? You might need him to pitch to one more batter…

Was hitting Chavez for Wilson the right move there? I think I’d rather have Wilson for an extra inning than Chavez for one AB.

I just can’t believe how poorly Washington has managed this WS.  Even my normal managerial standards.  I don’t know how Jon Daniels (the Rangers GM), who is somewhat numbers oriented, can stand watching these games?

"Was hitting Chavez for Wilson the right move there?”

Yes, of course.  You are down by 3 runs!  I don’t know that I would bat Chavez though.  Doesn’t he have anyone else?  I guess he figured LaRussa would bring in the lefty anyway, so Chavez was just a lame duck…

MGL/21 - Did you mean why hit for Carpenter?

He’d already gone 6 innings, and was leading off the bottom. You give up an awful lot of offense by letting him hit. And he’d been through the lineup nearly 3 times.

If he came up with 2 out and nobody on, I’d be much more likely to leave him in. But giving up the leadoff hitter is a big penalty…

See, that’s just it.  If TLR truly did detect some godlike power latent in Carpenter’s arm, he wouldn’t have yanked him so readily.  The idea that you’d watch Carpenter retire every single batter he faced...bizarre. 

Just pushing random buttons.

I shouldn’t have said, “Of course.” I think so. Not 100% sure.

If Andrus is so frustrated at flying out at 3-1, maybe he should take the 3-1 pitch?  Seriously.  Andrus versus Dotel is not a very productive hitter.  Probably close to a pitcher hitting level.  With Hamilton coming up, I think it might be a take.

If you want to know what an idiot McCarver is (not that you need any examples):

He has been criticizing the Cardinals for playing Freese even with the bag against Kinsler, because Kinsler never bunts.  Well, you idiot, he never bunts when the third baseman is playing in!  In the 7th inning, Freese was playing back. What do you think Kinsler did?  Which McCarver excitedly pointed out to us.

Moral of the story:  The third baseman has to play in such a manner that the expectancy from the bunt attempt and the non-bunt attempt are the same.

#25, I was being sarcastic.  Of course you hit for Carpenter…

MGL/28 - Whew.... I should have inferred that!

Pretty bad home plate umpire tonight!

That’s a wrap, folks!  Worst managed series ever? 

Here’s a task for aspiring sabers in the off-season:

Figure out around how much WE both managers cost their teams during the WS.  Seriously.

The not pinch hitting for Carpenter, and then pulling him so quickly in the next inning was flabbergasting.

But did you hear the interview with Larussa at the podium?  He said that it was Molina who told him that Carp still had plenty left, so that’s why he let him hit!  And yet he pulled him out at the first sign of trouble!  That’s Orwellian logic right there.

And yeah, MGL is right that Washington must not realize that a walk with the bases loaded is a guaranteed run.  (Well, we know he knows, but does he not know the risk/reward there?) To do that to your pitcher because, presumably, you think Freese is sooooo hot as to be the second coming of Barry Bonds is flabbergasting as well.  The Rangers got what they deserved in that inning.

And to the commenters above about Carp “not having it”: how predictive was your opinion?

I’m glad you stuck your b@lls there and posted as you did.  You provide evidence.

Everyone is a great forecaster after the fact, but lousy before the fact.  And the reason is simple: people are humans, and not machines.  Don’t look for patterns in humans.

The point is that it’s extremely difficult to figure out how persistent someone’s “on-ness” and “off-ness” is.  We need thousands of games to find 10 or 20 points of wOBA gap.  It’d be like knowing that Matt Holliday hits better against same-side pitching by watching him hit against 100 pitchers.  These splits, these patterns, are so subtle, that only mountains of data can find it.

So, next time you think you know if someone is on: you don’t!

And to the commenters above about Carp “not having it”: how predictive was your opinion?

I’m glad you stuck your b@lls there and posted as you did.  You provide evidence.

Everyone is a great forecaster after the fact, but lousy before the fact.  And the reason is simple: people are humans, and not machines.  Don’t look for patterns in humans.

The point is that it’s extremely difficult to figure out how persistent someone’s “on-ness” and “off-ness” is.  We need thousands of games to find 10 or 20 points of wOBA gap.  It’d be like knowing that Matt Holliday hits better against same-side pitching by watching him hit against 100 pitchers.  These splits, these patterns, are so subtle, that only mountains of data can find it.

So, next time you think you know if someone is on: you don’t!

I stand by my statements, because I’m not an idiot who takes a handful of innings as a legitimate sample size. I still think he was “off” last night (and man, I’m not God’s gift to scouting, I could be wrong, but note I didn’t say anything about him being ‘on’, which I’ve never claimed to be able to see), but of course that doesn’t mean he’s going to bsolutely get destroyed. Next time you want to make an inflammatory claim, you should really read what you’ve said (or at least come willing to clearly repudiate yourself):

The Book Says:
You can indeed tell if a pitcher is off based solely on the results of the first nine batters he faces. Such pitchers will perform somewhat worse than their true talent levels the rest of the way.

The Book Says:
If a pitcher is getting hammered early, there is a huge carryover effect for an inexperienced pitcher. For an experienced pitcher, there may be some evidence of a carryover effect.

"The Book Says:
If a pitcher is getting hammered early, there is a huge carryover effect for an inexperienced pitcher. For an experienced pitcher, there may be some evidence of a carryover effect.”

Obviously Carp falls into the latter category, and I don’t know that he was getting “hammered,” but your point is somewhat well-taken.

And yes, absolutely, the fact that he might not have “had it” that night, does not mean that he can’t pitch another effective 4 innings or so.

Which is another reason why the issue of pitchers having it or not having it, based on results, is really silly.

“And the reason is simple: people are humans, and not machines.  Don’t look for patterns in humans.”

I would qualify that to apply principally to professional athletes who are so well-trained that they typically do not vary in true talent very much from day to day.  More importantly, whatever variation in true talent there is is going to be dwarfed and masked by the much larger random variation around the true talent.

I know some people with think or say, “Well TLR knew that Carp wasn’t ‘off’ so he left him in!”

Well, we have already shown in those other (fun) threads, that when managers leave their starters in after pitching brilliantly for 5, 6, 7, and even 8 innings, those starters do NOT pitch any better than expected given their overall stats, and that is in thousands of innings of data.  So apparently they CAN’T tell when a pitcher is on (or off).

(I hate debating things that are very unlikely to be true.  Some people say, “Year, but it COULD be true,” and I say, “Yes, you are right, it COULD be true, but the evidence says that it is unlikely to be true.”

If I say, it is NOT true, that is only semantics.  Most of the time I say, “The evidence ‘suggests’, or, “It is likely to be true,” etc., but sometimes I tire of having to qualify everything by stating that we don’t know for sure.  That should be the standard disclaimer for all conclusions based on empirical data and should NOT have to be repeated every time!)

And for the record, I don’t see Tango’s post as being inflammatory or being directed at anyone in particular.  It was just a lighthearted attempt to point out that several people were posting how “off” Carpenter was and then he pitched 4.1 more scoreless innings. Just an anecdote to illustrate something he and I talk about a lot (that all of the pundits should put their money where their mouths are before and not after the fact)…

Maybe it was another thread where many people were calling Carp off… Above I saw just 2 comments to that effect: WanderingWinder/7 and MGL/11. Even the latter is debatable, as MGL said he ‘… is just not “on“‘, but if we’re in a binary world, not on equals off.

FWIW, WanderingWinder suggested in the first post that it might make sense for teams to hit for the pitcher each time, and go to the bullpen. So that would make him lean to pulling Carp early irrespective of his first inning issues.

Trev/2 speculated about how much Carp might suffer for pitching on short rest, but that came before the first pitch, and was not a reaction to the bad start.

I agree with Tango/33’s points, that it’s much easier to be right in retrospect than when predicting, and that a short sample in game is probably not indicative of anything. But if lots of people were posting that Carpenter was off and should be pulled early because of it, it wasn’t in this thread.

I wasn’t being inflammatory.

I did say “We need thousands of games to find 10 or 20 points of wOBA gap.”

And, everytime we do our studies, when we say that you can tell that someone is on or off, it’s because that’s the amount of evidence we do find.  That’s the extent of on and off.  It’s not that if you are Carp, that “off” means that he has “NO COMMAND”.  When we say “off”, we mean that you lose about 20 points of wOBA effectiveness.  And, you can’t visually see that someone is off, if you are trying to compare a .320 wOBA pitcher to a .340 wOBA pitcher.

To imply “NO command” would mean some ridiculously bad pitcher, a .400 or worse wOBA pitcher who has no business pitching.

And, it’s perfectly fine to say “I observe the pitcher to have NO command”.... but that still doesn’t mean much.  You can observe him to pitch him as crap, say he’s pitching at a .450 or .500 wOBA clip.  But all that means is that a true .320 wOBA pitcher will pitch at something like .325 or .330 the rest of the way.  That observation is not so actionable as to require quick hook.

Geoff, I was kidding when I said that he was not “on.”

I have not read the book but are you saying it says to pinch hit for the pitcher every time he comes to bat?

And what do you think were Washington’s mistakes in the series? I understand intentional walks are a disaster.

And I understand some of the logic about not being able to tell if someone is hot or cold. With that logic should Kohi Uehara been left on the WS roster? He was a disaster in the playoffs and the latter part of the season. But clearly its not big enough of a sample size. How would you guys handle a situation like that, keep pitching him?

The biggest part of the reason that I want to pinch hit probaby every time for the pitcher was because it’s game 7 of the World Series. So the book isn’t going to say anything there - it only happens once every handful of years. Hasn’t since ‘02. And the reason here is that you have really good relievers to bring in all the time, with all your starters available as relievers (with decent bonus for relieving). And the PH is going to be much better. So I mean, the first time up, depending on how early it is in the game, levverage situation of the PA, it’s going to be questionable. But you really ought to PH for the pitcher a lot, and not leave bullets in your gun.
Carp being off was simply more evidence for the maybe possibly close first time through - and I was leaning toward thinking this was the right decision even if Carp’s spot came up in the 1st before the game.

I believe they talked about what goes into a good pinch-hitting decision in an earlier thread on World Series manager mistakes.

The gist of it is that if your starter is about to face these batters for the third time through the order, you should almost always pinch hit.

The reason is because hitters do better and better each time through the order, and the effect size is pretty strong, something like 30 extra points of OBP each time through the order. That late in the game, even a good starter isn’t going to be much stronger than a solid reliever (and you’re probably only going to get one more inning out of him anyway), but he is going to be a MUCH worse batter than a solid pinch hitter.

Also, I think mgl was mocking LaRussa for giving up the offense of a pinch hitter and then pulling Carp after he pitched to ONE batter; if LaRussa’s putting him on such a short leash, there’s no way LaRussa could’ve honestly believed it was worth giving up the extra offense.

Bill, I could not have said it any better!

LaRussa didn’t know that he would take Carp out after 1 batter but, as bill said, he was clearly on a short leash.  It doesn’t take a genius to figure out that if you are planning on taking your starter our after 1 base runner or 2.... Oh wait…

Uehara’s a very strong pitcher, with a strikeout-to-walk ratio over 7 (although he’s an extreme fly ball pitcher, which means his weakness will be giving up too many home runs, and Texas is not a good ballpark for him to be pitching in).

You shouldn’t pull a guy that strong unless you have a very good reason to believe he has stopped pitching like a talented pitcher, and 10 plate appearances (though he only got 4 outs and gave up 2 homers) isn’t enough to tell you anything by itself. While it’s possible for a guy to snap under pressure or to temporarily lose his stuff for psychological reasons, it doesn’t happen nearly as often as 10-PA runs of good and bad luck, so most streaks of 10 really horrible plate appearances are made by pitchers who will get up and pitch well the next time out.

So yes, the standard play based on the numbers is to keep him in and play him the same way you would play him if he hadn’t pitched to those 10 guys.

I might defer to a manager’s judgment when it comes to a pitcher’s temporary true talent level (not really, but I can give it lip service, can’t I?) IF that manager has demonstrated even a sliver of intelligence and rationality. Washington has not, so he DOES not even come close enough to sniff the benefit of the doubt.

Would you all ever say a batter can be “hot”?

Or David Freese is just getting lucky?

I’ve read enough to believe the IBB are horrible and Molina is a .300 so walking him is probably nuts.

But when do you determine a stat is meaningful? there are so many ways to slice and dice them. For example Torrealba was 1/27 as a pinch hitter. Does this mean anything to you?

What about people saying Feliz should never come in with people on base. He’s a much better reliever starting an inning. Is that noise or legit?

Perhaps you all would say a computer could run the team but someone has to tell the computer what stats are important (ie is men on base/not on base important for a reliever? is pinch hitting a different situation than batting regularly?)

And Uehera actually gave up 3 homers. 5 hits (3 homers), 2 walks, 1.1 innings

Newbie, it would take around 270 printed pages to answer those (and similar) questions. Fortunately someone has already written a book like that…

"but someone has to tell the computer what stats are important”

The smartest thing anyone has ever said with regards to computers and humans.  Well done!

The ONLY thing that a computer does is make a human process human-data faster by using human-created software based on human-specified parameters (i.e., “what stats are important").

If the “computer” is giving out crazy answers, then the breakdown is with the software or the parameter list, both of which are purely human-based.  That is, it’s the human’s fault.

i don’t have the coding chops for it, but i would really love to see fangraphs (or bref, whoever) implement a tool that could advise on pinch hitting, sac bunts, and/or intentional walks.

let you enter the inning/bases/outs/count? states.  then let you choose the pitcher and hitter (and on deck hitters if relevant) via a drop down or search.  run a quick regression to guess at the true talent levels involved, and then spit out a result, whether it be a number or a thumbs up/down.

its a bit of work, but i think it would be useful as a quick tool for saberists as well as for helping to demonstrate to non-saberists why a decision was good/bad.

"a bit” ?  That’s ALOT!  And asking for someone to do it for free is unreasonable.

Would you all ever say a batter can be “hot”?

Or David Freese is just getting lucky?

What is the implication of being “hot”?  What is the limits of knowing that someone has been “hot”?

newbie/46

Every single PA has A LITTLE BIT of meaning. And every individual, situational PA has A LITTLE BIT MORE meaning when making same-situation predictions (e.g. a single pinch-hit PA means a little more than a single non-pinch-hit PA for predicting how a guy does the next time he pinch hits).

The problem is, one same-situation PA is often only worth (in terms of how well it predicts a hitter’s future performance) about as much as 1.1 other-situation PAs. (That’s just a rough estimate/rule of thumb and the actual number depends on what situation you’re looking at.)

So Torrealba’s 1-for-27 would be “worth” about 30 regular PAs in predictive power, which is a lot less predictive than the 400+ total PAs he had over the course of the whole season. If you added those three extra PAs of 1-for-27 hitting to your overall prediction to account for that extra predictive power, you would estimate a batting average about 2 or 3 points lower than his overall batting average, which is not much of a difference.

Of course, players in general do worse when they pinch hit than they do otherwise. But the best way to predict (using just the numbers) how Torrealba will do as a pinch hitter is to take his regular performance and then subtract a penalty equal to how much worse the average player does in pinch-hits PAs as compared to non-pinch-hit PAs. That’s because you have 400+ PAs for Torrealba this season, and thousands of PAs to show you the real difference between pinch hitting and being a starter.

If you do that, you’ll probably wind up predicting that he’ll hit .220-.240 or something, rather than .037, which is what you’d get if you just took the 1-for-27 number and treated it as gospel. I think the .220-.240 estimate looks much more realistic.

There’s more mathematical detail about this stuff in posts that talk about “regression to the mean”; for me as a computer/math guy, the posts at 3-dbaseball.net about regression to the mean and beta distribution were especially helpful.

51/tango

as it turns out, it looks like someone has done half of this work already.  i think dave studeman gets the credit here.
http://www.hardballtimes.com/thtstats/other/wpa_inquirer.php?view=standard&runs=4.5&base=4&inning=18&outs=1&score=-1

the other half is still a lot of work, but i still don’t think it is unreasonable. the calculations are not super intensive, the difficulty would lie in linking to players in a database.

and sorry, but your “doing it for free” comment is to me a weak argument. ten years ago it would have to be hard to imagine that the wealth of data you can now get on bref/fangraphs would be free. if someone with the coding skill can get it done and drive more hits to their site, it might be worth the opportunity cost to them.

if not, we can continue to do these things by hand, as it has been done in these threads or in klaasen’s fangraphs post on pujols.

First off, that looks like my data with studes’ interface.  You call that “half” the work?  It’s nothing close to what you need.

Derek Zumsteg has an iPhone app that has something of what you need.  Whether that’s good enough, I don’t know.  You can try that.

And my argument isn’t “weak”.  There’s hundreds, if not thousands, of people qualified to do what you are asking, and they ain’t doing it.  Why do you think that is?

Just because someone HAS done something for free (and actually this is only mostly free), doesn’t mean that you can expect people to continue to do that thing for free in the future, much less do something more for free.

by “half” i meant half of my suggestion, with one part being the win expectancy situation and change and the other part being the addition players involved.  i did not mean half of the total work, as it is clearly the easier requirement.

when you said the doing it for free comment, i figured you meant that if someone were to do it they would offer it as an app or program to buy.  i didn’t think that was necessary, due to the wealth of information that many sites give us for free due to the wonders of advertising.

as for why no one has done it, that is either because they have A. not thought of the idea or B. didn’t think it was worth their time, as i referenced with the opportunity cost comment. 

if everyone of ability thinks choice B, so be it.  i am just putting option A out there. the world series was such a comical display with regard to ibb/bunt/sb that it seemed to me that many saberists (or even mainstream types) might have more interest in this information.

i understand that this is your turf and i am not trying to argue with you at all. i just want to explain my reasoning better. i love this site and have tremendous respect for you, i just didn’t like my thoughts being dismissed in a way that i thought was a little too quick.

55/wanderingwinder

sorry, i posted that before i saw your comment.  i agree, you can’t expect or demand anything when it is being offered for free. i was just tossing out the idea.  i probably underestimated the work that would need to go into it, but i think the world series demonstrated that such a tool might be useful.  if not for the cost, of course.