I forget who the story was about - Steve Carlton or Tom Seaver or someone - but I remember hearing that on the morning of a 17 strikeout performance (or some similar feat), the pitcher looked down at his arm during warmups in the bullpen and felt like it didn't belong to him, like it was some alien appendage.

And for one day, he got to experience the life of Mariano Rivera.

Like I said, they're humans. Sometimes we have bad days.

As an extreme example, it's pretty obvious that Rick Ankiel didn't have it the day he threw 42 wild pitches and wasn't going to get it if he kept pitching. (The next game he pitched, granted it was the next season, he struck out 8 in 5 innings and didn't throw a wild pitch.) Or conversely, Kerry Wood struck out 20 guys in one game. Obviously he was really on that day. There is no reason to believe that he can't also be really off another day. (Cue jokes about him being off for the rest of his career.)

Your response to someone bringing up a study and multiple followups is to bring up an opinion and two anecdotes? Not very convincing.

One thing I've always wondered: do pitchers really "have it" on certain days and "not have it" on certain days, or is this mostly random?

I have no doubt that pitchers can have it/not have it on certain days. I have considerable doubt we can identify the former from the results of the previous inning(s).

Not arguing, it just seems like the kind of thing that should "obviously" be identifiable, one way or another.
We have an incredible amount of extremely reliable raw data. It's hardly impossible to winnow out irrelevant factors. If we can't draw a conclusion, given the mountain of evidence available, my singular genius tells me we're approaching the issue incorrectly.

***

Btw, in general, when people write 'mgl has studied the issue and found x' that statement has no useful meaning.

Your response to someone bringing up a study and multiple followups is to bring up an opinion and two anecdotes? Not very convincing.

Do you think athletes are robots that perform the same way every game?

I have no doubt that pitchers can have it/not have it on certain days. I have considerable doubt we can identify the former from the results of the previous inning(s).

But if you can't identify it based on previous innings, then pitchers don't in fact "have it."
This is an issue where, if you can't find it, then it's not there.

Btw, in general, when people write 'mgl has studied the issue and found x' that statement has no useful meaning.

What does this mean?

Do you think athletes are robots that perform the same way every game?

Nope. They're inconsistent and have wide swings of performance. I certainly think it's possible that looking at weeks and months worth of data might be more predictive than looking at 3 innings. Even if those innings just happened.

I don't know for sure and I haven't even read the book. But if you're saying that a pitcher can have "it" for 3 innings and then lose "it", or not have "it" for 2 innings and then find "it", well the current state of having "it" or not doesn't really mean anything then.

It's worth distinguishing between "having it" and "not having it." There is some evidence that if a pitcher gets hit really hard in the early innings, he will continue to pitch worse than his usual performance. Presumably, in at least a few cases this is an indication of injury, and so it will be predictive. For a healthy pitcher, I don't think we really know if extremely poor performance is predictive. But there is really little evidence that pitchers "have it" on some days -- no matter how well a pitcher is throwing, his future performance does not seem to be any better than usual. Very counterintuitive, obviously, but that's what the data seem to show.

Nope. They're inconsistent and have wide swings of performance. I certainly think it's possible that looking at weeks and months worth of data might be more predictive than looking at 3 innings. Even if those innings just happened.

Then I'm not sure what you are arguing, because I didn't say anything that disagrees with that. I'm saying that sometimes guys just have a bad day from the first pitch. And, along with at least one other on this thread, that when they do, they usually get pulled before they can get deep enough into a game for it to show up in a study that doesn't account for that.

I don't know for sure and I haven't even read the book. But if you're saying that a pitcher can have "it" for 3 innings and then lose "it", or not have "it" for 2 innings and then find "it", well the current state of having "it" or not doesn't really mean anything then.

Well, someone else brought up that point. I agree with it, because again these aren't robots we are talking about. And I'm not sure what "doesn't really mean anything then" unless that's just another way of saying "if you can't describe what "it" means then it doesn't exist."

But there is really little evidence that pitchers "have it" on some days -- no matter how well a pitcher is throwing, his future performance does not seem to be any better than usual. Very counterintuitive, obviously, but that's what the data seem to show.

Wait, didn't we have a huge thread about very good SPs pitching the 9th when they were "on"?

And didn't the data show that these very good SPs had like a 2.50 ERA in the 9th when they were "on"?

That seems contrary to your statement.

But there is really little evidence that pitchers "have it" on some days -- no matter how well a pitcher is throwing, his future performance does not seem to be any better than usual.

I don't know why you think it should, since good outcomes don't track all that closely with good efforts in baseball. But I find the idea preposterous that there aren't days when pitchers (hell, all athletes) are not perforing closest to their maximum ability (and others when they're not).

Let's look at it this way, for simplicity's sake (in reality, there'd be shades of it - and most days a pitcher is probably neither in a state of it or not it - and varying degrees of results).

You can have:

It/Good Results in Innings 1-2
Not It/Good Results in Innings 1-2
It/Bad Results in Innings 1-2
Not It/Bad Results in Innings 1-2

In innings 3-4, the guys in the first group may very well continue to produce at a higher level and the guys in Column 4 (the ones who don't get yanked first) may well continue to pitch poorly (with results to match). Meanwhile, the middle guys start to get results that more closely line up with their effort. But if you're trying to track that simply by examining the results, it looks like there's no correlation between pitching well one inning and continuing to do so the next inning.

So, to me, the research doesn't disprove that pitchers can't have "it." What it demonstrates is that we can't identify what "it" is simply by looking at the results.

There is some evidence that if a pitcher gets hit really hard in the early innings, he will continue to pitch worse than his usual performance.

It's more than just being hit hard. Not locating the plate, not hitting top speeds on pitches also factors in.

Let's say a pitcher is a "true 3.50 ERA" pitcher. What kind of performance would we expect to see out of him from batter to batter and inning to inning? I know it depends on platoon/quality/etc of the opposing hitters (and tiredness of the pitcher), but wouldn't a 3.50 ERA pitcher still bounce around from batter to batter and inning to inning out of pure randomness centered around his base talent?

That seems to be what wee see in DMB, for instance.

We're not talking about the expected fluctuations in a game. We're talking about when a "true 3.50 ERA" pitcher starts off the game pitching like a "true 7.00 ERA" pitcher and doesn't improve as the game goes on. Just because someone is an athlete doesn't mean they can't have "one of those days" at work.

All of the evidence I've seen suggests this is largely or completely wrong

Go look at the PitchFX data at Fangraphs. Pitchers' average fastball velocities swing multiple MPH from game to game. If that's not evidence to suggest that pitchers' sometimes "have it" and sometimes don't, I don't know what is. That early game results are not necessarily predictive of late game results doesn't change that fact. A pitcher might not even pitch worse with his B stuff. Given that Walter Johnson completed about 5/6 of his starts and had a career ERA around 2, I'm guessing he was able to to adjust when he didn't have his best stuff.

It's one thing to take Johnson's statement and say, "OK, but so what? It probably doesn't really affect the pitchers' effectiveness all that much whether they have their best stuff or not, and we can't really do anything with this knowledge anyway," and a totally different thing to say, "All of the evidence I've seen suggests this is largely or completely wrong." To react the latter way is to be the saber Murray Chass.

What does this mean?

Oh, no snark intended. It's just when someone mentions a study that occurred, but without a link or salient details, we can't draw any conclusions no matter how illustrious the study's author.

Very counterintuitive, obviously, but that's what the data seem to show.

If anyone can link to the data, I'd appreciate it.

Here are the findings on how starters pitch in the late innings. It turns out, contra Snapper/59, that starters do not pitch any better in the 9th than we'd expect them to, even though they've pitched great for 8 innings (on average).

Re 60 and 64: if "having it" or "being on" doesn't actually mean a pitcher performs better, then I don't see how these concepts have any meaning. The claim becomes not only unfalsifiable, but also completely uninteresting. If Johnson pitched just as well when he didn't have the "good fastball," then who cares?

If Johnson pitched just as well when he didn't have the "good fastball," then who cares?

That Walter Johnson could still dominate without having his fastball is more a testament to his greatness than a debunking that players have bad days.

Here are the findings on how starters pitch in the late innings.

Ah yes, I remember that. It stems from this discussion and all its cross-posting. It was an... interesting thread.

That Walter Johnson could still dominate without having his fastball is more a testament to his greatness than a debunking that players have bad days.

I hope this is facetious. If the argument here is that "Walter Johnson had his good days and his bad days, but he was so damn great he managed to pitch just as well on his bad days," well, then, I really don't even know what to say.....

if "having it" or "being on" doesn't actually mean a pitcher performs better, then I don't see how these concepts have any meaning. The claim becomes not only unfalsifiable, but also completely uninteresting. If Johnson pitched just as well when he didn't have the "good fastball," then who cares?

I fully believe the pitcher does in fact get better results when he's on (and worse when he's off), but that is is not absolute due to the multiple other factors that influence a pitcher's results (luck, the play of his opponents, the performance of the umpire, etc.).

Take Ray's true talent ERA 3.50 pitcher. That can mean that on any given day, at his normal abilities (the level of performance at which he's most commonly found), his results can fluctuate between 2.50 and 4.50.

But when he's on, maybe his range of outcomes fluctuates between 2.00 and 4.00. When he's off, between 3.0 and 5.0.

So by looking at the results, we can't identify whether a pitcher was on/off/or somewhere in the middle (and one reason why one inning's performance is not predictive of next inning's). But our inability to identify a pitcher's performance level through the results doesn't mean it's not there, nor that it doesn't have an affect on those results.

Now, whether only that which can be proved about baseball is interesting to you, that's your call. But I find that perspective rather narrow for my tastes.

So by looking at the results, we can't identify whether a pitcher was on/off/or somewhere in the middle (and one reason why one inning's performance is not predictive of next inning's). But our inability to identify a pitcher's performance level through the results doesn't mean it's not there, nor that it doesn't have an affect on those results. Now, whether only that which can be proved about baseball is interesting to you, that's your call. But I find that perspective rather narrow for my tastes.

Sorry, but this is incorrect. If it can't be measured, it's not there. Sure, there will be a lot of noise, as you say. But if pitchers are sometimes "on," let's say 25% of the time (or whatever), then the days they pitch very well for 7 innings will consist of a disproportionate share of the "on" days -- perhaps 50-60% of the time. And because they are "on" much more than usual in these games, then they should perform better in inning 8 (not always, but on average). But, in fact, they do not. The only way what you're saying could be true is if the "on" effect is very tiny compared to other variations -- maybe a reduction in true ERA of 0.10 or something. And if so, I would ask again: "who cares?"

,

I hope this is facetious. If the argument here is that "Walter Johnson had his good days and his bad days, but he was so damn great he managed to pitch just as well on his bad days," well, then, I really don't even know what to say.....

Well you could start by accurately quoting me. I didn't say "he managed to pitch just as well on his bad days," I said "without having his fastball" which is not the same thing. Not having your A pitch but still having your B and C pitch on a given day is not the same as not having any pitches at all on a given day.

So by looking at the results, we can't identify whether a pitcher was on/off/or somewhere in the middle (and one reason why one inning's performance is not predictive of next inning's). But our inability to identify a pitcher's performance level through the results doesn't mean it's not there, nor that it doesn't have an affect on those results.

Exactly. We can't predict it, and it's not easy if at all possible to differentiate from a game where a guy is just unlucky. But it's a reason. Sometimes when a guy says "I just didn't have it today" after a lousy performance, it's not just a line. If we can't see that this stuff happens in sports then we're trying to hard to live in a black and white world.

Exactly. We can't predict it, and it's not easy if at all possible to differentiate from a game where a guy is just unlucky. But it's a reason. Sometimes when a guy says "I just didn't have it today" after a lousy performance, it's not just a line. If we can't see that this stuff happens in sports then we're trying to hard to live in a black and white world.

Whenever discussions of this type come up one always sees this type of statement repeated, and my question is always as follows: if your explanation is completely unfalsifiable and offers no predictive power, then why should be believe in it as opposed to magic pixies, angels in the outfield or whatever other crazy explanation we can come up with?

I'm not actually certain that this is a settled issue-just because we haven't found something yet, particularly when we haven't looked in all the right places yet, doesn't mean it doesn't exist. But the idea that regardless of what we find we're just going to keep believing in this, or that no amount of evidence would be sufficient to overturn this belief is troubling to me. Ultimately if explanations about guys "being on" or "having it" are correct that *has* to show up in the data somewhere or it's just magical thinking.

But if pitchers are sometimes "on," let's say 25% of the time (or whatever), then the days they pitch very well for 7 innings will consist of a disproportionate share of the "on" days -- perhaps 50-60% of the time. And because they are "on" much more than usual in these games, then they should perform better in inning 8 (not always, but on average). But, in fact, they do not. The only way what you're saying could be true is if the "on" effect is very tiny compared to other variations -- maybe a reduction in true ERA of 0.10 or something. And if so, I would ask again: "who cares?"

Or, if the actual on days are far fewer than 25 percent, which seems like an absurdly high percentage of the time to expect a pitcher to be at/near his optimal performance level.

I honestly don't care if this level of discussion doesn't interest you. But don't deign speak for the rest of us. Baseball interests me beyond that which can be precisely measured.*

*Or, that which can be measured at this particular moment. Perhaps someday an enterprising MGL-wannabe will figure out a way to isolate and measure this.

I honestly don't care if this level of discussion doesn't interest you. But don't deign speak for the rest of us. Baseball interests me beyond that which can be precisely measured.

Oh please. There are many, many interesting things in baseball that can't be measured. Unfortunately, in this case you've selected a theory which, if it can't be measured, actually isn't true. And I plead guilty to not being interested in factors that do not actually exist. PBAF said it well:

Ultimately if explanations about guys "being on" or "having it" are correct that *has* to show up in the data somewhere or it's just magical thinking.

I'm not actually certain that this is a settled issue-just because we haven't found something yet, particularly when we haven't looked in all the right places yet, doesn't mean it doesn't exist.

This I agree with. Like I said, if it's possible, it's not easy.

But the idea that regardless of what we find we're just going to keep believing in this, or that no amount of evidence would be sufficient to overturn this belief is troubling to me.

Well, I think it's foolish to automatically waive away what we can't measure and say it's because it doesn't exist. We can't measure work ethic, but we know it exists.

I think it would be an interesting thing to study with Pitchfx. Look at the starts where the guy has more zip on his fastball, or more movement.

For the numerous (and often justifiable) complaints about MGL's tone, he acquitted himself quite nicely here. Well done MGL.
Agreed.

Seconded.

Setting aside platoon issues and the like, what we really need to know is: What is the true talent level of my pitcher as he's standing on the mound.

There's a difference between a true 3.50 ERA pitcher giving up 3 runs in an inning because of randomness centered around his base talent level, and this same pitcher giving up 3 runs in an inning because he actually took the mound in that inning as a 5.50 ERA pitcher.

Flip a six-sided die 10 times and you might get 2261552224 -- but that doesn't mean the die is weight towards landing on 2.

Discussion is useless without a strict mathematical definition on "on"