Cliff Lee Was Not Living on the Edge
Sometimes a graphic can make the obvious even more obvious. Anyone watching last night knows that Cliff Lee did not look like the guy who took the Rays and the Yankees to school. A look at his pitch plots makes the reason clear.
First, against the Yankees, you’ll notice that Lee avoided a particular part of the zone:
There is that one conspicuous cutter sitting dead center, but nothing else comes even close to it. You see, for the most part, cutters in the bottom half of the zone, change-ups low and away to righties, and curveballs low in or below the zone. If you happen to find someone with baseball savvy who had not watched this game and showed him this strike zone plot, he’d probably be able to tell you that the pitcher had great success.
And then there was last night:
There are a number of concerning aspects of last night’s strike zone plot, not least of which is the number of pitches near the center of the zone. The top-right portion of the plot is also concerning. As you can see, those four curveballs had no chance. Just three of the 11 curves he threw were strikes, none swinging, while against the Yankees he threw eight of 16 for strikes, including one swing and miss. Last night he managed just two curveballs low, while he did it consistently against the Yanks.
What further hurt Lee was his lack of a changeup. Against the Yankees he threw it 14 times and got nine strikes, three of them swinging. Last night he broke it out just five times, and each time it came early in the count. Three were strikes, but none generated swings and misses. Each time the Giants swung at the change, they put it in play.
Finally, the cutter caused him some problems, too. In both games he kept the cutter mostly in the lower half of the zone, but the difference was in how he painted the edges. Against the Yankees you can see the black dots spreading pretty far to each side of the zone. Against the Giants there aren’t many cutters on the outer thirds. The Giants, unsurprisingly, put far more cutters in play, 22.6 percent, than the Yankees did, 15 percent, even though they whiffed more (16.1 percent to 10 percent).
No pitcher, not even Cliff Lee, can be perfect every time. After three incredible postseason starts, he finally had a game where he didn’t have complete control of his pitches. Sometimes aces can gut through starts like that. Other times they’re going to get hit around. The Giants had their moment in Game 1, but unless they can complete the sweep this won’t be the last they see of Lee. In Game 5 I’d expect Lee’s strike zone plot to more resemble his ALCS start than his first World Series one.
As an end note, I think Tim Marchman nails it with this parody quote:
“I didn’t have my good stuff going tonight,” he said. “But I doubt that made a difference. I’m not a six-sided die, but mathematically I act like one and function with surprisingly little agency. Any game I pitch is just an expression of my true talent, that of my opponents and something that isn’t quite what the average person means when they say ‘luck’ but works more or less the same way. I hope for a 90th percentile outcome every time out, but to me it’s really all about sticking near my mean outcome and giving the guys a chance to win.”
14
With that quote, Marchman redeems himself for the egg he laid on Slate, lol.
Haha…I love the Marchman quote. While the center of the strike zone did not repel his pitches, as it generally does, he still struck out 7 with just one walk. There was some dirty LOB% going on.
Which is one of the reasons I have an issue with FIP (or other pseudo-DIPS stats). FIP thinks this was a great game, when looking at the strikezone plot, you can see he was terrible.
Let’s not discount the plan that the Giants took with them to the plate. They did similar things against the Phillies with Halladay and Oswalt, but they really had success this time. I think we have to really credit hitting coach Hensley Meulens with a lot of the Giants offensive success this year, essentially getting the absolute max out of a lineup that shouldn’t be that great.
I didn’t have a dog in this fight until I found out that the Giants employ both Bam Bam Meulens and Roberto Kelly. Was Claudell Washington’s phone # unlisted?
You don’t have a dog in this fight? You mean a dog in this race, right? Mike Vick, is that you?
The storyline seems to be that the Giants were up there hack-hack-hacking, but they were actually very patient. Too patient even, letting a number of right-down-the-pike fastballs go, and getting called out on strikes FIVE times, four against Lee, and often on very hittable pitches. Lee looked rattled after the second inning.
Even with Lee “missing his spots” he had 7k, 1bb in 4.2ip. The Giants had a .533 BABIP vs. Lee with an .866 SLG. Wouldn’t this kind of night be dismissed as a fluky outlier all year long? Or is this an argument for pitchers negatively influencing BABIP? I can’t stand the gray area!
When a lot of balls are left up in the zone, BABIP and SLG will be higher, right?
Pretty much.
That’s why BABIP is an eye-roller.
Did you watch the game? If so, then you saw that the Giants were hitting the balls hard. When the ball gets hit hard, it’s more likely to be a hit than when it doesn’t get hit as hard. And hits that are hit hard are more likely to become extra-base hits than hits that are not hit as hard. The Giants were hitting the ball hard because Lee did not have good location and many of his pitches lacked movement. Please, enough with the BABIP nonsense.
I think it is a bit of both. The giants DID hit a bunch of balls hard. However, hard-hit balls sometimes go right at a fielder — not with SF batting last night. key up-the-middle fliners in the 5th could have been right at a fielder, and stopped the rally there. On the flip side, did you see the several hard-hit balls that went right to andres torres in the late innings? The double-play ball to end the top of the 2nd was a smash as well.
brendan:
I agree, hard hit balls often get hit right at fielders. But I never said that hard hit balls always become hits, I said they are “more likely” to become hits. By the same token I would never say that weakly hit balls always result in outs, because they don’t. It’s a matter of probability. The harder the ball is hit, the faster it goes. The faster it goes, the less time a fielder has to react. The less time the fielder has to react, the closer the ball has to be to him for him to make a play. If you realize this, then it becomes clear that whether a ball in play results in a hit is not simply a matter of random luck. Rather, in the aggregate, the probability that it results in a hit should rise as the ball is hit harder. Of course it is not a perfect correlation, since, for example, weakly hit ground balls will have a higher probability of becoming a hit as the runner’s speed increases. But even in the latter situation, only very weakly hit grounders usually result in hits for very speedy runners. So I speak generally. If one accepts that a pitcher has a large amount of control over the type of contact that is made, then one should accept that a pitcher has a large amount of control over the probability that a ball in play becomes a hit. And if one accepts this notion inning by inning, game by game, one should not rationally conclude that the lack of variation in BABIP over long periods is due to random luck.
You can’t call it a fluke when that many pitches were right down the middle
Soooo…you guys are saying that Cliff Lee is NOT a robot?
Cliff Lee missing his spots is mostly due to randomness isnt it? The analysts seem to imply its some kind of skill lost overnight when that is most certainly not the case. Its not like Cliff needs to overhaul his overall approach.
Sort of. The randomness is more likely in not knowing which game a pitcher’s “stuff” is present.
“Skill level” sometimes is used like assuming a pitcher’s ability is a static constant from game to game. It’s not. It’s an averaging across his games.
Knowing in which games a pitcher’s “stuff” is present is not random. You may analyze incorrectly, but that’s not random. Please stop randomly using the word “random.
Besides, knowledge of when a pitcher’s stuff is present is generally increased by a combination of watching the game and knowing what to look for. It cannot be determined by looking at a boxscore.
Human action, such as aiming a baseball and throwing it, is not random.
Well, but its random in some way I guess. Its like a basketball player shooting free throws. The skill is indicated by the percentages, but inevitably a player is going to miss on random occasion, but you cant control whether thats happening during World Series Game 7 or in a regualr season game.
But it’s not random. It’s mechanical variation. If a robot is designed to imitate a certain action, then it will, unless it breaks or it has a flaw. All jokes aside about Cliff Lee or Roy Halladay, we all know that humans are not robots. Even at our best we cannot reproduce the same exact action over and over. So if it looks like a pitcher threw the same pitch in the exact same way twice in a row, we know better. Just because we can’t easily measure the difference with a handy statistic doesn’t mean it is random.
Yes, I agree it’s a mechanical variation, but as you say yourself, this kind of variation is hardly in the realm of the pitcher. Obviously good pitchers seem to have less mechanical variation or at least theyre able to successfully overcome it. Nevertheless this variation is going to appear on a random basis from a statistical point of view.
Mechanical variation is “in the realm of the pitcher,” and I did not say, or mean to imply, otherwise. If it weren’t, then there wouldn’t obviously exist better pitchers with less mechanical variation.
Your argument is flawed. You’re arguing that from a statistical point of view, there is no discernable pattern, therefore the variation is random. The truth is that the mechanical variations are not random, and so neither are the statistics. Statistics do not cause what happens on the field. What happens on the field is what gets reflected in the statistics.
True, what happens on the field is what gets reflected in the statistics. I totally agree. But if you project the future success by deriving it from past games then you have to assume those variations occur randomly since there seems to be no clear cut reason as to why a variation appears in a certain point in time.
Ah, yes. Statistics will not tell you anything about mechanical variations in pitching delivery, and so if we want to predict when such variations occur based entirely upon statistics, we have to assume that the variations are random. But that is very different than saying, as you said earlier, that “Cliff Lee missing his spots is mostly due to randomness” or “you [the pitcher] cant control whether thats happening during World Series Game 7 or in a regualr season game.”
Its conceivable that Lee didnt change his approach (since there was no apparent reason to do so given his performance this post-season), thus we may assume that he would eventually regress based on his overall career stats, meaning he was “bound” to miss his spots sooner or later. Eventhough he had actual control of his pitches those misses were random in some way.
If by “approach” you simply mean pitching mechanics, then I agree, except for the part where you say “those misses were random in some way.” No, the misses are not random in any way. We can think of possible explanations, such as cumulative physical fatigue, cumulative mental fatigue, increased anxiety or whatever. Just because we don’t know the reason or can’t predict when it will occur does not make it random. It just does not follow.
Peoples belief that what they are unable to explain is random variation is much like religious belief that what we can not explain must be due to divine intervention. The devil lies in such assumptions.
Cliff Lees poor performance was due to poor execution of his pitches. He had no curve ball to speak off and his FB lacked command.
Why that is nobody knows. Maybe he had a bad breakfast, a fight with his wife, a bad hair day, a sore arm, not enough sleep, too many off days, didn’t like the SF mound, balls were too slick, or just bad luck. We can;t say for sure and who cares?. But just because we do not know or can not say for sure what it was that causes a poor performance does not mean it it is random event.
If you blow a big sale that should have been a slam dunk because you were 1 hr late for an appointment, try to explain to your boss it was just random variation and bad luck but that your true talent level remains the same and good luck is on the horizon as you will regress to mean.
Somehow I agree with you guys, maybe it’s just that we have different definitions of what constitutes “randomness”. To further my point I refer to one of Dave’s (Cameron) articles called “Accepting Randomness”.
Quote: “These opinions are generally held because of the outright refusal to accept randomness. The idea that something could happen repeatedly, without cause, is very hard to for a lot of people to swallow. But it’s true, and it’s a very important concept to buy into when trying to project the future performance of baseball players. Random happens. ”
Here’s the link:
http://www.fangraphs.com/blogs/index.php/acceptingrandomness/
pft –
Thank you. A voice of reason.
Mr Romo’s clavicle –
Perhaps you do have a different definition of the word “random” (though if you do, I would argue that it’s the incorrect definition).
However, the article you quote betrays the fact that some people, including Dave Cameron, do not have a different definition of the word, and yet still believe that a pitcher’s performance can be pure randomness. Specifically, his use of “without cause” to define random behavior indicates this. Sorry, but human behavior is not random. I’m not sure whether to laugh or cry when I read something that equates human behavior to a coin toss.
Its about thinking in terms of probability instead of the usual deterministic (black/white, true/false, success/failure) way. Joe DiMaggio’s hitting streak demanded a great deal of skill, but ultimately its “luck” (randomness), if somebody is able to string together a 56 game hitting-streak.
The probabilities are not fixed. Pitchers and hitters can do things that raise or lower the probability of success. Is there luck involved? Yes. But the skill involved is so much greater than the luck involved that calling the whole process “luck” or “random” is disingenuous.
There was a Cliff Lee pitch chart presented here at FG earlier, that to my amazement showed that He lives in the center of the zone (early) and up and away (later).
Given our perception of him as a non-power pitcher it went against common thinking.
Urine had an at bat in the game where he took 2 pitches right down the middle and then swung at a curve in the dirt, a very good example of a batter getting himself out.
I believe the data presented on Lee previously was for the entire season, meaning that he generally has success pitching in the zones that were similar to G1 of the WS.
I would have guessed that he pitches more like Tom Glavine than Curt Schilling, but he doesn’t.
For every guy that “misses his spots” and gets rocked, there’s another guy that misses his spots and gets 5 lazy flies and another 5 balls fouled off. Isn’t that the whole point of the BABIP debate? The Giants scored 2 or less runs in 55 games this year. Are we to believe the opposing pitchers in those 55 games just had really good location that night?
Good location is only one of many factors that a pitcher can control. Pitch movement, pitch sequence, changing speeds, pitching to a hitter’s weakness, pitching to the defensive alignment, and so on, are others. How these factors relate to one another and are combined in many different ways has a lot to do with the many different results that you see in each game. It’s not a convenient truth for BABIP fans, but that doesn’t make it any less true.
Another factor, of course, is the quality of the hitter. It’s no secret that the Giants did not have a very good offensive team this year. The worse the hitter, the lower the standard on the pitcher to do something to increase his chances of getting a favorable result (and vice versa). It’s no surprise that the Giants scored so few runs so many times, and it doesn’t undermine my argument. My point is that all other factors being constant – batter, ballpark, etc. – the things pitchers control from the mound have a much larger impact on the result than BABIP fans give them credit for. If you vary the batter and leave the pitch constant, that will obviously have varying results as well. Another unfortunate truth is that during baseball games, the pitches vary and the batters vary. It gets complicated, but that doesn’t make it random.
Its conceivable that Lee didnt change his approach (since there was no apparent reason to do so given his performance this post-season), thus we may assume that he would eventually regress based on his overall career stats, meaning he was “bound” to miss his spots sooner or later. Eventhough he had actual control of his pitches those misses were random in some way.
If by “approach” you simply mean pitching mechanics, then I agree, except for the part where you say “those misses were random in some way.” No, the misses are not random in any way. We can think of possible explanations, such as cumulative physical fatigue, cumulative mental fatigue, increased anxiety or whatever. Just because we don’t know the reason or can’t predict when it will occur does not make it random. It just does not follow.
Does (2B+3B/AB) allowed vary in the same way as BABIP?
I don’t know the answer, but it seems like that stat would measure how much a pitcher is getting “hammered” (outside of HR allowed, of course).
Such a stat could help, in that a 2B or 3B is usually hit relatively hard. But if the pitcher allows only singles, it doesn’t follow that he’s not being hit hard. In other words, you can’t conclude that a pitcher’s BABIP was unluckily high just because he didn’t allow a 2B or 3B.
Also, I don’t buy the argument that HR’s are a random percentage of fly balls and therefore should be treated differently…
Isn’t the point that if you just checked the stats, then C.Lee had a good game by FIP and so fWAR (albeit not pitching very long), 4.2IP, ~1.92BB/9; 13.5K/9 7:1 K:BB, 0HR, just unlucky with the BABIP and the extra base hits
So to project him forward, we chalk it down to a bit of bad luck, poor defense etc, and still like him going forward (which would seem to be correct as we would like Lee pitching for our team at a palatable price)
Then we look at how he played and see that he threw more over the heart of the plate than normal, so maybe deserved some of what he got.
FIP and fWAR as projection for future gains, is fine and useful…used as a system to reward past performance, (i.e awards) not so good, as Lee would have been rewarded for this performance, when it wouldn’t be justified.