Let’s Now Be Critical of a Single Pitch Selection
The only pitch that should literally never be thrown is a pitch aimed at a hitter’s head.
Anything else, totally fine. You don’t read MGL over the years without learning some things about game theory. Game theory explains that, optimally, you need to be unpredictable. You should bunt just often enough so that your opponent doesn’t know if you’re going to bunt. You should pitch out just often enough so that your opponent doesn’t know if you’re going to pitch out. And you should mix your pitches just enough so that your opponent doesn’t know what pitch will be on the way. It’s simple, if oversimplified: don’t tip your hand. It does your side a disservice.
Game theory is fascinating, and at the same time analytically limiting. When you get to talking about pitch sequences, any pitch, in isolation, is justifiable. Any pitch should/could be thrown more than zero percent of the time. Let’s say there’s a hypothetical that calls for, I don’t know, 60% fastballs in, 39% changeups away, and 1% hanging sliders. That describes no real situation, but anyway. If you see the pitcher throw a fastball, okay, yeah, that should happen sometimes. If he throws a changeup away, same deal. And if he throws a slider down the middle? It seems like a mistake, but every so often it does make sense to do that on purpose, in theory, because otherwise the hitter could just rule the pitch totally out. When a pitch gets totally ruled out, it slightly tips the balance. Part of being unpredictable is the willingness to sometimes do things that don’t seem so good. Surprising mistakes can be surprising successes.
Because of game theory, it’s almost impossible to reasonably criticize any given pitch or pitch sequence. A pitch comes with an n of 1, and stripped from context, you don’t know how many times that pitch would’ve been thrown in the same situation. Taking one pitch and only one pitch, you almost always have to conclude that, maybe it was fine. There’s no such thing as a pitch that absolutely should never be thrown, aside from the one noted at the beginning. This is frustrating, but sometimes sensibility frustrates. So the world can be.
And yet. I think this is against my better judgment, but there’s a pitch I want to criticize. It happened in Wednesday’s Game 2, and it was thrown by Hunter Strickland to Salvador Perez. I can’t declare absolutely that the pitch was a terrible idea, because of all the reasons, but this is about as close as I can get to believing that a pitch shouldn’t have been called. Perez, against Strickland, broke the game open. He did so against a pitch I think he knew damn well was coming.
Setting: bottom of the sixth, 3-2 Royals, two on, one out. Strickland comes in from the bullpen, to face Salvador Perez. Perez homered in Game 1, but for the most part, for a while he’s been a mess. The game here is still within reach. Strickland is known to have a high-90s fastball, and a couple other pitches he doesn’t throw much. He never throws changeups to righties anyway. From the start, Perez knows he’s going to see fastballs or sliders.
Pitch 1: fastball, away, over the plate. It’s fine. Perez swings.
Strickland’s going to start most plate appearances with fastballs, but sometimes he won’t do that, and most hitters don’t like to be too aggressive on the first pitch, especially if they haven’t seen a guy before. So after the foul, Strickland’s ahead 0-and-1. Makes a breaking ball more likely. That means there’s also value in throwing a heater. Strickland places another, tucked in the low-away corner.
Beautiful pitch! It got Harold Reynolds to say pretty nice things. It put Perez in an 0-and-2 hole, and Perez has gotten pretty used to that by now. There were options. Fastball, or breaking ball. A breaking ball would probably be low and away. A fastball could perhaps be up, but in Strickland’s appearance in Game 1, all his two-strike fastballs to righties were intended to be around the outer edge, at the level of the knees. That’s how he retired Lorenzo Cain, and that’s how he tried to retire Josh Willingham before dropping in a pretty slider.
And this is where things turned. The call, unsurprisingly, was for a breaking ball low and away. The Giants know that Perez is ultra-aggressive. The idea was sound, but the pitch was too bad — too bad for Perez to chase, and too bad for Buster Posey to block. The runners moved up, meaning one was suddenly within 90 feet. A runner 90 feet away can score on another bad breaking ball in the dirt.
The count was still 1-and-2, heavily in Strickland’s favor. But this is where Perez’s brain would’ve kicked in. From the beginning, he would’ve eliminated the changeup. Now, he would’ve figured he could eliminate the breaking ball. Strickland wouldn’t want to risk it again, and Posey also wouldn’t want to call for a pitch Strickland wasn’t controlling. And Strickland likes his fastball, even as a putaway pitch, so Perez had plenty of reasons to think he was going to get some heat. And where does the heat go? Perez had seen it away, between the knee and the thigh. It’s what Cain saw the night before. It’s mostly what Willingham saw the night before. It didn’t seem at all likely that Strickland would try to throw a fastball up in the zone. The one time he did that to Willingham with two strikes, it was a location mistake.
Perez had every reason to expect a fastball away, somewhere around knee height. The Giants should’ve recognized that. The count was still very much in their favor — they couldn’t let Perez have an edge in the guessing game, not with his back up against the wall. Yeah, it would be hard to call again for a pitch that Strickland had just bounced, but what would be the odds of another bouncer? What would be the odds that Posey wouldn’t be able to block it? And there was another choice, too: a fastball located somewhere else. Perez is used to getting pitched down and away, in particular with two strikes. He kind of hunches over in his stance, as if to better reach those pitches away from him. So that would’ve been another choice: fastball up. Maybe Perez gets it in the air deep enough to score a run, or maybe he swings and misses or swings and pops up. There should’ve been three options. It seemed like everyone agreed there was one.
Strickland did not miss by very much. The ball did drift a little over the plate, but it was still on the edge, and besides, Perez is capable of hitting pitches outside. Look at the way he attacked: he was leaning over, looking for a pitch in that spot. For his career, Perez has hit better than .300 against such pitches, and those were presumably in less predictable situations.
The hitter shouldn’t be able to know what’s coming. He especially shouldn’t be able to know in a 1-and-2 count. I’m not saying Perez knew with absolute, 100% certainty, but he would’ve had a real good idea. And the Giants should’ve known that. Based on what Posey called for, Strickland didn’t miss in a surprising area. He barely even missed at all, relative to the norm. Strickland mostly executed the pitch, but the pitch wasn’t a good pitch to execute under the circumstances.
“After he threw that breaking ball to Salvy, he didn’t want to throw another one in the dirt,” Royals third baseman Mike Moustakas said. “I’m pretty sure Salvy was hunting dead red and got a pitch, and he smoked it out to center field.”
This is kind of basic stuff. Hitters anticipate pitches all the time. But guess hitters should be exploited in pitcher-friendly counts, whereas in this case I feel like Perez and Posey were on the same wavelength. I don’t know that for sure, I haven’t talked to Posey or Strickland, but I feel like the odds of such a fastball in that case were very high. Too high, given that Perez would be hunting for the same thing.
Back to game theory. Because of game theory, we can’t say anything about this absolutely. It was a sample size of 1. It’s always a sample size of 1. But let’s say this were repeated 100 times. How many of those times would Strickland throw that kind of fastball? How many of those times would he throw a different fastball, or another slider? My guesses are too often and not enough. Sometimes hitters are just going to be far too comfortable. That’s when game theory says you throw them a curveball. Figuratively, or literally.
Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.







Neck pitches, totally fine.
Ah, so many things for me to say, where to begin…
Your basic premise is correct, and I appreciate the hat tip.
Then the entire rest of the article goes against your very premise, which was initially correct! That’s like when you read (or hear), “Small sample alert, but..” the writer or speaker doesn’t need to finish the sentence because we already know it has no validity (because it is based on a laughingly small sample size), but he does anyway. You did the same thing!
That being said, your statement:
“There’s no such thing as a pitch that absolutely should never be thrown, aside from the one noted at the beginning.”
Is not true. First of all, using a what we call a GTO (game theory optimal) approach by a pitcher is only optimal if the batter is using a GTO approach as well. If he isn’t, then the pitcher still uses an optimal approach (of course – that is tautological), but we don’t call it GTO AND it is difficult to determine precisely AND it may (and likely does) change as two, related and co-dependent, things happen: One, your opponent’s stratgey changes for various reasons, two, you force your opponent to change his strategy by virtue of your non-GTO strategy. Keep in mind though that GTO is always a good strategy even if your opponent is not doing the same (in which case your GTO strategy is not really optimal against his non-GTO strategy – you can do better). No one can ever exploit a GTO strategy by definition. If you don’t use a GTO strategy, and this really applies to baseball and pitching, because you think your opponent is not using a GTO one and therefore you think your strategy against him is better than a GTO one (which it should be if executed properly), you are running the risk of being exploited. Using non-GTO strategies can be dangerous and must be done very intelligently and carefully. I’ll get more into that in a second.
You said (you probably think that I said it too) than any pitch (or any possible strategy if we are not talking about pitching) is available to be thrown in any given situation even if it has a very small non-zero probability of being thrown. That is not correct.
Game theory merely says that you use all possible alternatives in such a way that your opponent can never take advantage of your strategy. There is nothing they can do to get an edge. That leads to the win expectancy of all of your strategies yielding exactly the same results. That is the important point. The actual (not the observed which contains all kinds of random fluctuation) result in win expectancy (run expectancy is usually an OK proxy) should be the same for all pitches thrown in any given situation (not for that batter or pitcher overall).
So, it is possible that even if yo throw a certain pitch 1% (or .1% or whatever the threshold at which the batter virtually never can expect it) of the time, it’s value is still worse than some other pitch that is thrown 100% of the time. If that is the case, then you eliminate that pitch. That comes up occasionally. For example, with a 3-0 count against a bad hitting pitcher, it is likely that for most pitchers the value of the fastball and probably the fastball down the middle, even when expected 100% of the time (certainly if it is correct for the pitcher to take 100% of the time), is better for the pitcher than an off-speed pitch even if it is never expected because of the possibility of missing the strike zone with the off-speed pitch even if thrown down the middle.
Anyway, like I said, it is not true that all of your alternatives must be available in any given situation, only that whatever your alternatives are, if you use them in the proper proportion, they all should have the same expected value, assuming that the opponent is using a GTO strategy.
If he isn’t, then all bets are off, and the proper strategy becomes difficult to figure out and complex.
But, your initial point remains true – to some – degree. If a certain alternative is part of your strategy and your strategy is optimal, then you can NEVER be sure that a pitch was thrown in error as long as there was a non-zero probability of it being thrown, which, as I said, is not always the case (but usually is).
On the other hand, Bayes will tell us that that if we observe something which was only supposed to occur a small percentage of time, there is a much greater likelihood of it being a mistake then if it was supposed to occur more often. Sort of like if you think your kid stole the cookies but you are not 100% sure, should you yell at him or punish him (that is really the essence of our justice system, no?). A little bit of a philosophical question.
Anyway, I have to run now, but I want to get back to the rest of your analysis later in light of what I just wrote…
READ THIS!
Seriously, don’t think tl;dr here. Read it, and take the time to read it slowly.
Jeff wrote a great article, and as usual he put in a lot of caveats about the difficulty of analysis.
But if you like to get additional insight on the analysis of baseball, you owe it to yourself to understand how game theory applies. What MGL is describing here is not nitpicking or getting off on being confrontational by any means. Rather, he’s laying out many of the practical considerations in trying to apply game theory to analyzing pitch selection.
Well worth the read. Like a free bonus section to “The Book”, IMHO.
Sorry I don’t really know what kind of authority MGL is on game theory so maybe I’m slightly disagreeing with the PhD of game theory, but I must object to one sentence (yes nitpicking): “No one can ever exploit a GTO strategy by definition.” In this part I have to side with Jeff Sullivan. If I KNOW my opponent will always follow GTO, I can use that to my advantage. In card games this is possible all the time. If another player consistently follows the same strategy, then essentially you have more game data than he does and more data is an advantage.
I believe the GTO strategy is a reference to the optimal strategy that results in a Nash equilibrium. As MGL says in his 5th paragraph, this is only optimal when all the players adopt the GTO strategy, resulting in an equilibrium. By definition, if you know your opponent is going to follow a GTO strategy, your optimal strategy is a GTO strategy.
In most real-life games, except for tic-tac-toe and the like, there is not a clear or easily recognizable optimal strategy, so it’s rare to actually have a single true GTO strategy. And if you have more information than your opponent, the game-theory equilibrium falls apart and you have an exploitable advantage, at least until your opponent spots how you’re exploiting it and adjusts his strategy to decrease your advantage.
Jim, you’re not disagreeing with the PhD of game theory, you’re going a step beyond that and disagreeing with the definition of the terms. By definition a GTO strategy CAN NOT be exploited. If you play a card game using a GTO strategy no amount of information in the world overcomes that fact (except of course for non-strategic information, e.g. physical tells in poker). A major part of GTO strategy in most games involves some amount of randomness. That is to say, even if you know your opponent to be playing GT optimally you will still not know what they are doing; they will (by definition) be randomizing so perfectly that you can’t know what they are doing (unless there is nothing you can do about it, in which case they don’t need to randomize).
Using poker as an example: If you play sub-optimally, you may bluff too often or too little. If I am aware of this I can call more or less. However, if you are playing optimally you will be bluffing and calling in the exact proportion that it does not matter if I call or fold. Even if I know exactly which strategy you are using I can’t do anything to take advantage of it.
A similar baseball example would be a situation in which a fastball or a slider are equally useful to the pitcher, with the pitcher having no other pitch worth throwing. The pitcher, if he is being GTO, will throw a fastball 50% of the time and a slider 50% of the time. If the hitter knows that he will get fastball 50% and slider 50% . . . he still has no idea which one he’ll be getting on this pitch! So there is no way to exploit the information, even though he knows exactly the pitcher’s strategy. If the pitcher is using a non-GTO strategy, perhaps throwing fastballs 60% of the time, the hitter can expect a fastball with greater frequency to gain an advantage.
Another way to think about GTO strategies is that they are strategies that do not lose their effectiveness if you tell your opponents what you will be doing. That is how perfectly non-exploitable they are.
This brings up an interesting question, for me: is a pitch that the pitcher can execute but has such a bad platoon split that it typically isn’t thrown to right-handers a worse choice than being locked in to a pitch the batter is expecting?
It seems to me that if everybody knows he won’t throw the changeup it’s the perfect time to go against that knowledge.
The problem is that there is a reason that he doesn’t throw the changeup: it’s an easily below average pitch with little movement that he has very little command of. If you throw a pitch like that down the middle, hitters will hit it even if they are not looking for it. If you throw that pitch way out of the strike zone, hitters will not swing at it.
Ok, let’s parse the rest of the article although I can already tell you are going to be “results blinded.”
“Makes a breaking ball more likely. That means there’s also value in throwing a heater.”
First sentence is true I assume. While he probably throws 70 or 80% fastballs on the first pitch, he probably throws 50 or 60 or 70% (I have no idea) fastballs on 0-1 counts (depends on the hitter, score, runners, outs, inning, etc.). The reason you throw more pitches, especially ones you cannot control as well on pitcher’s counts is simply that the ball is not as costly and the batter protects the strike zone more.
Your second sentence is a throwaway sentence. He throws fewer fastballs at 0-1 than 0-0. Period. The actual pitcher is (presumably thrown randomly around those percentages, like you throw a pair of dice in which 7 comes up 1/7, 12, 1/36 etc. Pitchers and catchers don’t “think” they throw randomly around a certain percentage, but they do. At least the good ones do. They think that there are thoughts that drive the exact decision and to some extent that’s true, but for all intents and purposed the exact pitch is randomly determined or decided (of course human beings are not good an randomizing things – just ask people to write down random lists of numbers and you will see much a more choppy list – fewer “sequences” – than a truly random set of numbers).
So what does, “That means there is also value in throwing a heater” mean? Again, the value of the heater and the slider, if those are the only two pitches he throws in that situation MUST be the same – by definition. If they are not, he is throwing the one with more value not often enough. That is common sense. If I decide to throw fastballs and sliders 50/50 in any given situation and the value of my fastball is greater than the value of my slider, in that situation, then I must throw my fastball 51%. If it is still more valuable, I must throw it 52%. So and so forth until the values are equal!
“There were options. Fastball, or breaking ball. A breaking ball would probably be low and away. A fastball could perhaps be up, but in Strickland’s appearance in Game 1, all his two-strike fastballs to righties were intended to be around the outer edge, at the level of the knees. That’s how he retired Lorenzo Cain, and that’s how he tried to retire Josh Willingham before dropping in a pretty slider.”
Ok, let’s talk about this a little. Sliders to same side hitters are almost always thrown down and away. Virtually always. Occasionally you see a slider thrown inside meant to break into the zone, but that is rare – really rare. I’m talking about intent of course. And sometimes you see a “get me over slider” either on the first pitch or on fastball count. 0-2? 98% of the time, it is coming down and away to a same-side batter.
Opposite side batter is different. Sliders are generally back-doored outside to outside corner, or down and in (back foot slider). Of course throwing a slider to an opp-side hitter is dangerous unless you have a really good one. Anyway, this is a same-side situation.
Keep in mind that the reason the slider is going to be thrown down and away if it is thrown is because all the other locations, even if the batter does not know it’s coming have a worse value (for the pitcher) than any other pitch if they batter knows it is coming 100% of the time. In that situation.
The fastball can have many locations depending on the batter, situation, pitcher, etc. And each location can have a zero or a non-zero chance of being thrown to.
“The count was still 1-and-2, heavily in Strickland’s favor. But this is where Perez’s brain would’ve kicked in. From the beginning, he would’ve eliminated the changeup. Now, he would’ve figured he could eliminate the breaking ball.”
That last sentence is insane. Why would he “eliminate the breaking ball,” leaving only the fastball? With a runner on third base, the chance of the slider probably goes down because it is a less effective pitch because of the chance of a wild pitch. But a zero chance? It could be, but highly unlikely.
And the idea that the batter eliminates a pitch so then the pitcher can throw it with impunity makes zero sense from a game theory perspective, unless the batter is an idiot and the pitcher is a genius, which I suppose is possible. Major league catchers are really smart. That is why so many of them become managers.
If it is correct for Perez to eliminate the slider THAT MEANS THAT IF THE PITCHER THROWS IT EVEN THOUGH THE BATTER DOES NOT EXPECT IT AT ALL, IT IS THE WRONG PITCH! By definition, if the batter can correctly eliminate a pitch that means that if it is thrown
it has LESS value than his other pitches even when the batter is completely surprised.
I am going to stop right here. The rest of your article about the pitch selection uses the same incorrect logic and failure to understand how game theory works.
I hope I have shed some light on it. The batter/pitcher confrontation is NOT about one trying to outguess the other and one trying to fool the other. Whatever the pitcher thinks, the batter can think the same thing and vice versa. The antidote to that is game theory optimal approached by batter and pitcher. I am not saying that they all are perfect in that but they are pretty darn close. Certainly the best batters and hitters are.
“the idea that the batter eliminates a pitch so then the pitcher can throw it with impunity makes zero sense from a game theory perspective”
Is this what Jeff Sullivan is saying? I don’t think this is what he’s saying. Isn’t he just saying that, in his own personal mind, Salvador Perez has concluded that the breaking ball is not coming back? It certainly is possibly that
1) Perez thinks that there’s a 0% chance he gets another breaking ball
and ALSO that
2) There is a >0% of Strickland throwing a second breaking ball.
These aren’t mutually exclusive.
“By definition, if the batter can correctly eliminate a pitch that means that if it is thrown
it has LESS value than his other pitches even when the batter is completely surprised.”
I don’t understand this at all. Let’s say Strickland has a third, I don’t know, the perezball, which he never ever throws in game for some reason or other. Nonetheless, this pitch is absolutely nasty and would get hitters out 99% of the time if he threw it. Surely then only 1) here is true:
1) the batter can correctly eliminate a pitch
2) it has LESS value than his other pitches
I’m no expert on game theory so enlighten, these are just my thoughts
At some point there is a split between practical and theoretical. I suspect the pleasure MGL derives from baseball is not the same one you or I do.
Shermham, I believe that MGL’s post was assuming that the discussion was about a situation in which both players are performing optimally, or at least somewhat close to it. Your post raises legitimate points if one or both players are playing extremely sub-optimally. Normally we can assume batters and pitchers, being so highly trained, are performing somewhat close to optimally, although I suppose either of these two might be the rare exception.
MGL, with all due respect, I think you may not completely understand game theory with regards to optimal pitch frequencies.
GTO means that you have reached an equilibrium in which if you move in either direction, your pitch selection will be more exploitable – that is if the batter has full knowledge of your pitch frequency. This is a grand assumption which is probably never completely true, so GTO isn’t perfectly applicable to pitch frequencies.
However if you are attempting to figure out some GTO frequency, the goal is to maximize the total pitch value of all pitches, NOT equalling the pitch values of fastball and slider as you have suggested.
Against someone like Perez, it is possible that optimal frequency is something like 60% breaking balls and 40% fastballs. In this case, fastballs may actually have a negative pitch value while breaking balls have a highly positive one. By your logic, this would suggest that you should increase the frequency of throwing breaking balls until pitch values are even, which is completely incorrect.
I’m assuming you have enough background in mathematics to know how derivatives work. At the optimal pitch frequency, the derivative of the total pitch value is zero because the pitch value is at a maximum. I am going to use fastball frequency as the independent variable X, but you could also use breaking ball frequency and arrive at the same result. So you can express our equation we are trying to solve as DT/DX = 0, T being the total pitch value and DT/DX being the instantaneous rate of change of total pitch value with respect to fastball frequency. Since total pitch value is the sum of the breaking ball’s pitch value and the fastball’s pitch value, you could also express this as DB/DX + DF/DX = 0, where DB/DX is the instantaneous rate of change of the breaking ball’s pitch value with respect to fastball frequency and DF/DX is the rate of change of the fastball’s pitch value with respect to fastball frequency. Subtracting DF/DX from both sides of this equation you get DB/DX = -DF/DX. So what this is saying is that at the optimal GTO pitch frequency, if you move an infinitely small bit away, you will be losing the same amount of pitch value in one pitch as you are gaining in the other. This is NOT the same thing as saying that both pitch values are equal! (If you were to move a non-infinitely small bit away, you will be losing more pitch value in one pitch than you gain in the other)
I’m assuming this makes perfect sense to you and the detailed explanation wasn’t needed, but I just wanted something for other potential readers to see with regards to what GTO pitch frequency actually is.
TL;DR version: At GTO pitch frequency, the pitch value of the fastball plus the pitch value of the breaking ball is at a maximum. The pitch value of the fastball does not likely equal the pitch value of the breaking ball at this optimal frequency. One could actually be highly negative and the other highly positive, but it would still be GTO.
A small error on my part – I started the post saying “slider” which morphed into “breaking ball” halfway through. I am referring to the same pitch in both cases.
Your penultimate sentence:
“Sometimes hitters are just going to be far too comfortable. That’s when game theory says you throw them a curveball.”
Is NOT game theory at all, at least the proper execution by an expert.
I freely and happily admit that it appears I still don’t understand game theory very well. Perhaps I shouldn’t have brought it up at all! The way I *feel* is that, in this particular instance, the actors were not acting optimally. It’s obviously impossible to prove, and all I have is a gut feeling and a Moustakas quote and some observations of player tendencies, but I think, in Perez’s head, he did all but eliminate the breaking ball, given what had just happened. (His discipline, keep in mind, sucks.) That’s not optimal, but that’s my suspicion. And that’s why I think he was vulnerable to another breaking ball that presumably would not have been as bad the second time around. But Posey and Strickland agreed on a heater, perhaps because the wild pitch was too fresh. I figure that’s also suboptimal, but Strickland appears to be stubborn and that was the one pitch he’d thrown well.
I could be way, way off. I think I included plenty of caveats. I don’t like to criticize these things, because the interplay is so complicated. But I feel more comfortable about this than I usually do, because I felt more strongly that the players involved weren’t acting in an optimal fashion. It felt conspicuous, but maybe I’m forcing things.
Jeff, the important takeaway, which you actually started with, and correctly so, is that there is rarely (and certainly not on an 0-2 count!) ONE pitch that is correct.
There are many potential reasons that they threw the fastball there. The most likely one is simply that it was one of two possibilities, probably with roughly equal percentages. Run that scenario 100 times, and Strickland is going to throw a fastball (probably in different locations some sub-percentage of the time) 50% of the time and a slider (probably down and away near 100%) 50% of the time. So there is NO right or wrong pitch. There can’t be on an 0-2 count. Given that, and it is almost a mathematical certainty, you have no beef whether he threw the fastball (in any location) or the slider.
But you want us to believe that YOU think Perez thought 100% fastball because there was now a runner on third and Strickland just threw a wild pitch in a slider? That’s preposterous. Perez is a catcher on a WS team. Do you think that HE thinks an 0-2 count in that situation requires a 100% fastball? Is he an idiot?
And if he genuinely thinks that, then surely the pitcher and his catcher would likely think the same thing and would know that Perez would also think the same thing. Do you see where I am going here?
That’s simply not the way MLB pitching and hitting work. And if it did occasionally, with certain batters, pitchers and catchers, you picked an awful example, with a catcher at the plate (who presumably thinks exactly like the other catcher, Posey) and a very good catcher in Posey.
You’re just way off base here I am afraid and it is difficult for me to explain why.
As I said, the best way to explain it an think of it is that whatever the batter, certainly if he is a catcher, thinks, the pitcher and catcher can and do think the same thing. And each one can also think of what the other one might think. That leads to only one thing and that is roughly game theory optimal approaches. If it didn’t, you would not be long for MLB.
Have you ever been a professional baseball player or talked to many? Because I am friends with multiple and, as a collegiate pitcher myself, have talked to them about pitch selection. You might be surprised to learn that they don’t think nearly as much about it as you seem to be purporting here.
But don’t let reality stop you from denigrating Jeff’s work in an unnecessarily abrasive way.
How I critique someone’s work is none of your damn business. I’m pretty sure Jeff can take it. If not, he can speak up for himself. He doesn’t need a baby sitter or guardian angel.
I’ve already said – perhaps you missed that part – that most pitchers and batters don’t think they are using game theoretical approaches and randomized decisions. All of them are pseudo-random anyway. As well, a person’s thought processes often appear “random” from the perspective of another person. That is more likely what goes on in a pitcher/batter confrontation. There is danger in that of course – the pitcher or batter can co-opt that non-random thinking and use it to his advantage. That likely happens a lot in baseball.
I’ve consulted for MLB teams for over 10 years and worked exclusively for one for 2 years. I’m not pulling this stuff out of my ass.
“How I critique someone’s work is none of your damn business. I’m pretty sure Jeff can take it. If not, he can speak up for himself. He doesn’t need a baby sitter or guardian angel.”
It’s not about whether he can “take it.” I’m sure many people could “take” a punch to the ribs, but it doesn’t mean that the fighter isn’t an asshole. Your abrasion adds nothing to the argument, yet you feel compelled to keep it. That’s you being an asshole, which has nothing to do with me being a “baby sitter” or Jeff being able to “take it.”
I’m glad that you have experience in baseball. In what capacity did you serve (if this won’t out you)? Did you consult over pitch sequencing or something similar? I’m happy to re-read your comments and I’m happy to admit if I’m wrong, but it seemed like you were making a strawman.
It is not about being right or wrong Paul. I don’t care if I or you are right or wrong. Those are generally subjective terms anyway. I speak the truth, and I am always intellectually honest. At least those are my goals. If anyone catches me in error on either of those I will be the first one to admit it.
If someone chooses to ignore what I have to say because they don’t like my tone, that is on them, not me. I am not here to please anyone. I am here to speak my mind and share my experience and knowledge. I do it the way I do it. Sometimes I am patient and considerate. More often I am not. If someone doesn’t like my online personality for whatever reason they are free to ignore my comments, although I personally feel like that is a foolish way to approach learning and life in general. I am not running for office or entered into a popularity contest.
I worked as an analyst for a team for 2 years. It is no secret. I have consulted for several others over the years. In those capacities I act differently because here I don’t give a hoot who reads/likes/learns from me and there I have responsibilities and get paid. Make sense?
Paul, did you realize that Jeff mentioned MGLs work in his piece? I think this at least gives MGL every right to critisize and correct the piece. And I’m prety sure Jeff welcomes that.
“If someone chooses to ignore what I have to say because they don’t like my tone, that is on them, not me.”
–said every @sshole ever
(Not saying that MGL is one, but people use that to give themselves license to do all manner of dickish things. “I’m just keeping it real” or “I’m just being honest!” Naw, you’re just a dick.)
“If someone chooses to ignore what I have to say because they don’t like my tone, that is on them, not me.”
–said every @sshole ever
-said most geniuses too
I think you are actually agreeing with the article, if I am understanding you and the article correctly. Sullivan is saying that Perez is not using a game theory optimal strategy by guessing fastball low and away and choosing to ignore other possibilities. This is not necessarily a poor strategy if you consider that maybe Perez believes he has no chance of doing anything productive with a slider even if he is looking for one, or if he believes that guessing fastball low and away just gives him the greatest chance of doing something productive. In this case, as you stated, Strickland and Posey should consider deviating from their game theory optimal strategy, and not throw a fastball in that area.
I thought he should have come in on his hands with the fastball. I was surprised they stayed away like that, and if they were going to throw that fastball away, it should have been more outside, off the plate, near the LH batters’ box.
Fastballs way off the plate away almost never get swung at. They are truly waste pitches than rarely get used. You will see high and away pitches thrown but not too far out of the zone.
I agree.
Posey set his target down and away. I can tell by Salvador Perez’s stance and swing that he is vulnerable on pitches down and away. I haven’t looked at any zone maps, but I am confident the data will back me up. Perez’s stance, approach, and swing are geared toward killing pitches up and away and down and in. In essence, he is a typical right handed hitter – albeit one who swings at a lot of pitches off the plate.
Granted, most hitters not named Trout or Cabrera are vulnerable on pitches down and away. But I cannot fault a decision by Posey and Strickland to follow up a badly missed breaking ball down and away with a fastball down and away. Strickland was utilizing his strength against Perez’s weakness. A properly executed fastball at the knees or a bit lower and away has a high probability of beating Perez, even if Perez is expecting that pitch and location. Strickland simply missed his spot by a bit. While a fastball high and in on the hands was a viable alternative, Perez is such a great high ball hitter that I can understand a decision to keep the ball down and away. It’s the safer pitch.
Strickland just missed his spot a bit, and Perez killed it. But it wasn’t a bad pitch. It’s the Major Leagues; even good pitches get hit hard.
Right, the only way you can keep a hitter swinging at those breaking pitches off the plate down and away is to sometimes throw a fastball in the same location. That is probably what happened, the pitcher missed his location a little, Perez happened to square it up and bang.
That’s why Romo threw that fastball to Cabrera at the end of the 2012 WS, even though his slider was lights-out that season.
thats my point. You dont necessarily have to get a swing and miss on that pitch. Its a 1-2 count, so you have a couple of pitches to play with. You could waste a pitch that sets up for the rest of the AB, at least change the hitter’s eye level or try to make it more difficult for them to figure out whats coming. The outside pitch I suggested wouldnt have been my first choice. As I said, I would have come in on the hands with a fastball. However, the reason I suggested the pitch should have been more outside in the event of throwing a fastball out there in that count, is that pitch likely doesnt get hit hard. As pitchers say, if you’re going to miss, you want to miss competitively, in an area where you wont get beat by making a mistake. In this situation, Salvador Perez was the hitter at the plate. I think its plausible that he expands the zone based on what we know about him. If he doesnt, its a 2-2 count. Again though, I would have come in on his hands with the fastball
Leaving aside the game theory aspect, which I know nothing about, the statement that Perez “could eliminate the breaking ball” seems a semantics issue. I took that statement to mean what Moustakus said, that Perez was likely hunting dead red, the kind of thing a lot of players say they do.
It’s kind of like saying a player gives 110%. That’s a silly statement at face value, but only a silly person takes it at face value.
Why does anyone care what Moustakus thinks Perez was thinking? Is Perez a mute?
Maybe he was for whatever reason. And maybe Strickland was going to throw the FB only 20% of the time, in which case Perez’ strategy (looking for the FB) was a bad one, but he happened to get lucky because he got that 20% pitch. In the long run, he is going to be badly fooled 80% of the time. Reason 1,281 why results based analysis is usually not very fruitful.
I think one of MGL’s main points over the course of this discussion (feel free to correct me if I’m wrong), is that good hitters, which is to say virtually all non-pitcher big-leaguers, operate under some semblance of Game Theory Optimal…that is, even when they are “sitting dead-red”, there is some part of their brain still active enough to identify and react to an off-speed pitch. Furthermore, a hitters relative skill is dependent on being able to “calibrate” these responses, whether consciously or not.
Some of the greatest hitters (Tony Gwynn and Stan Musial, for two) simply read and reacted. Of course, they both went up to the plate each time with a good idea of how the pitcher was going to try to beat them. But Gwynn hit the tits off of Greg Maddux, and never struck out in 107 plate appearances. Does anyone really think Gwynn (or anyone) could out-game Maddux?
I suspect the hitters who read and react outnumber the guess hitters in the Major Leagues.
This is probably a good chance to plug Coursera’s Game Theory MOOC. I’m a few weeks in and have found it profoundly interesting. Very challenging, but very interesting. I highly recommend it for anyone remotely interested in the field.
This post reminds me of the Princess Bride poison scene. Paraphrase: “I’m going to throw this pitch now because he expects me to throw a different pitch because no one would ever throw such an obvious pitch!”
If I’m reading the comments correctly, it appears that the third sentence of your article is also wrong.
Just going to leave this here, about a different, recent situation.
source
Had San Francisco catcher Buster Posey called for a breaking ball in that moment, Strickland said later, he would have shaken his head. Posey did not, freeing Strickland to nod, rear back and fire a string of unbroken heat to home plate.
[…]
“I know that I’m not going to back down,” Strickland said. “I’m just going to go with what I’ve got, give it my best and see how it goes.”
Clearly, Strickland wants nothing to do with game theory!
Part of the pseudo-randomization process, from the perspective of the batter, is the back and forth between the catcher and pitcher.
As I said, catchers and pitchers don’t think they are just flipping a 2 or 3 or 4 sided coin to determine pitch selection. In fact, they would think that the idea of a random selection around a fixed distribution in any one situation is ludicrous. But, whatever their thinking process that goes into a certain pitch – one pitch and one pitch only – essentially becomes random. At least for the better pitchers.
I feel like even though its obvious that a pitcher will generally stick with his best, its a little strange for Strickland to straight out say that he is going to throw fastballs