premium blogad:
blogads:
advertisement:
sitemeter:
|
|
|
March 6, 2007
Know Your Pitchers
If for some reason (like you’ve been in a coma) you haven’t checked the Greatest Website Ever Created—also known as Baseball-Reference.com—lately, well shame on you, because over the past few months site creator Sean Forman has been beefing up the site considerably, from the addition of splits, to the stat neutralization feature, to the new Play Index, which allows you to search for any single event in baseball history. The latest improvement is expanded pitch data for every pitcher and hitter where it’s available. There are quite a few missing plate appearances, especially in older data, but it’s still an extremely welcome addition to an already phenomenal website. Among these stats you’ll find: pitches per plate appearance, percentage of strikes thrown, first pitch strikes, swinging strikes, strikes looking, foul strikes, strikes put into play, called strikeouts, and swinging strikeouts. Furthermore, you can see how often a pitcher has seen the two most extreme counts, 3-0 and 0-2, and how that pitcher reacted. Let’s take a look at the Mets’ ace, Tom Glavine, a man with 290 wins and two Cy Young Awards on his resume. I find Glavine to be fascinating, a truly unique pitcher. He has the worst K/BB ratio when compared to the league average (8% lower) of any modern-era pitcher with 200 or more wins, save for two knuckleballers (Charlie Hough and Joe Niekro) and George Mullin, who is off the charts (23% lower!). He doesn’t let homers beat him, but he doesn’t get a ton of groundballs, either. Quite frankly, Glavine has achieved a tremendous amount of success despite being a guy who doesn’t strike anybody out and who doesn’t have great control. Even after watching him for four years, I still have no idea how he does it. Perhaps the pitch data will illuminate something. Maybe Glavine jumps ahead in counts, enabling him to get batters to swing at bad pitches late in the at-bat? Take a look at his career numbers (where data’s available) for how often his first pitch is a strike, how often he gets ahead 0-2 and how often he falls behind 3-0: 1st% 0-2% 3-0% Glavine (Career) 54% 13% 5% Average Pitcher 57% 16% 5% It sure doesn’t seem like it. He’s a bit worse than your average pitcher, and certainly doesn’t look like a Hall of Famer. Glavine’s a pitcher who lives and dies with his changeup. He certainly messes with a hitter’s timing; it might be safe to assume that a pitcher that does this will have a higher percentage of his strikes result in foul balls, either from the batter being way ahead of the pitch or far behind it. Unfortunately, the theory doesn’t hold water with Glavine—his lifetime percentage of strikes resulting in foul balls is only 26%, slightly below average again. Even when you look at the individual pitches, it’s still not clear how Glavine does it. That said, the data isn’t useless. One thing that is clear is that Glavine is very good at adapting. Glavine has kept his percentage of strikes thrown remarkably consistent: almost every year he’s around 60-61%. What’s especially interesting is how he gets them. Earlier in his career he probably had a couple miles-an-hour extra on his fastball, making his change look even slower. This resulted in 17-19% of his strikes resulting from swings-and-misses. Nowadays, that number has shrunken to the 10-12% range. Instead, he has gotten craftier with location; he’s made up for it in called strikes which have risen by 7-8% since his early career. He’s not throwing more strikes, but more of them look like balls to batters. This makes sense since the anecdotal evidence says Glavine has started using both sides of the plate more as he’s aged. Let’s move on to another pitcher who the Mets will be depending on substantially, namely John Maine. Maine certainly doesn’t have as much data as Glavine, so the small sample caveat definitely applies. Maine did a lot of things well in 2006, and the pitch data reinforces that. He threw 61% of his pitches for strikes, and he threw them early; 61% of his first pitches went for strikes, and he got ahead of batters 0-2 20% of the time. His stuff was also solid enough that he got his share of swings-and-misses, also. The homerun is the one problem Maine had last year, allowing 15 in 90 innings of work. Some of that is just bad luck, but some of it is likely due to pitch economy. He rarely gave up a homerun before the fourth inning (just two), which could mean either that he was getting fatigued or he was falling into easily recognizable patterns. The former seems likely: Maine threw 4.14 pitches per plate appearance, the highest in baseball of any pitcher who threw as many innings as he did, save Jon Rauch, a reliever. Hopefully Rick Peterson is stressing pitch economy to John this spring. If he does, a substantial drop in his homerun rate should follow. Finally, let’s take a look at Oliver Perez, the most intriguing pitcher on the Mets’ roster. Obviously, Perez enjoyed a tremendous amount of success in 2004 that he hasn’t been able to live up to in the years since. The easy culprit is his control, which has never been great, but it was at least acceptable in 2004 and hasn’t been since. Let’s take a close look at what the pitch data suggests. Year St% 1st% StL% StS% StF% StI% 3-0% 0-2% 2004 64% 59% 24% 20% 32% 24% 6% 22% 2005 60% 55% 27% 16% 30% 26% 6% 19% 2006 62% 59% 27% 16% 30% 27% 7% 20% The first and last two columns should be familiar from before, but the middle four represent a breakdown of what the batter did with the strikes Perez gave him: looked, swung and missed, fouled off, and put into play, respectively. Unfortunately Baseball-Reference does not separate the pitch data for Perez before and after his trade to the Mets, so we’ll have to keep things broad for now. The data illustrates exactly what Oliver’s peripherals already suggest: he threw more strikes in 2004 and blew more of them by hitters. There are some interesting things you can see from the data, though. In 2006 Perez improved his strike percentage by 2%. Doubtless, much of that came following his trade to the Mets where he only walked 17 batters in 36 innings. It’s somewhat encouraging. Second, he started throwing strikes earlier in the count again. He threw strike one just as frequently in 2006 as he did in 2004 after a nosedive in 2005, even if strike two still hasn’t come around quite as often as it did. Finally, he seems to have eased up on his power some since 2004—he may be throwing fewer strikes, but more of them are called strikes and fewer are coming as a result of swings-and-misses and foul balls. Furthermore, more of his pitches were being put into play. We know his velocity was down at times in Pittsburgh and I’ve heard it said that management there tried to turn the fireballer into more of a finesse pitcher; these numbers may lend credence to that thought. Getting back some of his aggressive nature can only help. In the end, how useful is this pitch data? I don’t know. I’d really need to pour over the data to see how predictive this stuff really is. While it sure doesn’t help explain Tom Glavine a whole lot, it may help other pitchers. Eventually, we may be able to find comparable pitchers with similar profiles that may be more reliable than traditional statistics. There probably isn’t enough data yet, but in a few years it may help us track players developmentally, providing a better idea of what a pitcher may become. 16 Responses to “Know Your Pitchers”Leave a ReplyYou must be logged in to post a comment. |
|
Nice article, but I find it hard to believe that these numbers mean much of anything. Talking about how Perez “improved his strike percentage by 2%,” you’re talking about less than two pitches a game. The most glaring drop-off for him from ‘04 are 3 or 4 swinging strikes a game. I don’t think you can pull anything meaningful here from the noise.
As much as I love numbers, I think a great comparison for sabermetrics is weather forecasting — whether or not it rains next Saturday isn’t really “luck,” but there are so many variables interacting, the vast majority of them unknown, that there’s no way to get a good prognostication beyond a 7-day forecast, and there never will be, no matter how sophisticated the analysis becomes. As famed meteorologist Henk Tenneken said, “We stand in front of the limits of predictability.”
Or, as a million people have said, “That’s why they play the games.”
Doesn’t mean it’s not fun to dig in to the numbers, I just doubt how fruitful it can be.
It is intersting to me, though, that Perez has the highest 1st% of any pitcher mentioned. Better than Glavine, better than league-average, better than Maine. The other numbers are the best, too — so why does he end up with ERAs over 5?
Positively fascinating material. Correct me if I’m wrong, but it sounds like the data from all games are treated equally. Is it possible to divide them up by wins, losses, and NDs? I realise that those divisions have as much to do with what happens after a pitcher leaves the game and with other forces beyond his control, but that would be part of the point. That is, often we complain that a pitcher did well but suffered from poor run support. Can this data help verify that pattern? Or does the data show other patterns, too, relative to wins and losses. Does Glavine pitch differently when he loses than when he wins, for instance?
Yeah I was thinking similarly to what Erick was thinking….except instead of breaking it down into Wins and Loses maybe break it down by Runs Allowed ie. does he throw more first pitch strikes in games he allows less runs?
By breaking the data into sections, we might be able to figure out if hes doing anything differently in the games he’s pitched well then the games he doesnt pitch well.
I’m not sure you can either, at least not without a lot more research than I’ve done. I was expecting more variation in the data. For instance there’s “only” a 13-15% difference in strikes thrown between pitchers with bad control in their worst seasons and pitchers with great control in their best (think Victor Zambrano and David Wells). You’d think there’d be more of a difference between one of the best and one of the worst. Interactions between slight differences in the data (all of which might not be presented on B-R.com) could have a larger effect, but who knows?
Maine actually had a better 1st strike percentage last year. In the end, I think it’s not all that important. Different pitchers will be able to succeed with different mentalities. Having some room to work with doesn’t necessarily mean you’re going to use it.
It is possible if you go into the data piece by piece, but right now it’s not organized that way. Interesting idea, though.
I do think it is odd and discouraging that Perez has seemed to lose his fastball from 2004. If you look at highlights from back then he was throwing 95-98 mph. Now he seems lucky to reach 92. Despite this I keep hearing from the Mets and those that cover them [Darling, Eddie Coleman, Cohen,etc] that he has “great” stuff. I guess I have several questions:
1]Why has he lost so much velocity at his age?
2]Were the Pirates trying to make him a finesse pitcher because of this or was their tinkering a cause of the velocity drop or is their an injury they knew about?
3]Can he recover the velocity?
4]Without the great heater and the way he currently throws, does he still have “great stuff”? Or did he used to have great stuff and now people are just saying he has “great stuff” because he used to when in reality he does not [see Zambrano, Victor]?
5]I guess I really just want to know why the hell he does not throw hard anymore?
I realize some of these questions might not have answers, but feel free to comment on any or all of them.
Even given their limitations, I think these numbers are perhaps the most interesting I’ve seen in a long time, since, they begin address the fundamental statements we make about pitchers — he throws strikes when he needs them, he gets behind batters too much, and so on. I suppose that’s part of your point, Alex, that the data do not back up the old saw that one pitcher is more successful than another because he throws strikes earlier in the count.
And humans and apes have about 99.9% of their genes in common. Numbers don´t tell the whole story, and tiny differences in the right spot can have huge repercussions. I think Glavine has a great sense of what the batters are expecting, and never gives it to them. His stuff may not be good enough to get strikeouts, but the batters still don´t hit it squarely enough to be effective. One millimeter makes a big difference when you´re trying to hit a sphere with a cylinder when both are moving with eccentric rotation and different velocity vectors. (How´s that for jargon?!?) That´s why hitting a baseball successfully is the hardest task in all of sports.
You have a flaw in your assumptions. Glavine has superb control.
By “control” I mean throwing the ball where he wants it. His game is throwing “pitcher’s pitches”. Down and away. Down and away. Down and away. Pitches on the black or just off the strike zone are very difficult to hit hard. Check his BABIP.
Glavine is wily enough to know that he’s better walking a guy than giving him something to hit. His game is wearing away at a batter’s patience because they just don’t know if the pitch will nick the black or fade 1 inch off.
Excellent data mining here. Though I agree with Tim, in that I don’t know if it tells us much. For example, if we are talking 2 or 3 pitches a game, it seems to matter who the pitches were thrown too. Getting the pitcher to miss is different than getting Pujols. That said, the count probably matters just as much to the batter, ala DW and his 0-2 prowess.
In addition, it doesn’t account for intentional balls off the plate and the like.
Excellent article though!
to not too. It is still pre-coffee time here in Cali.
I think these breakdowns are great for understanding how pitchers wind up throwing a certain number of pitches and pitching a certain number of innings. And, fo course, the stuff’s fun to look at. For analyzing effectiveness, though, I don’t think they’re really going to tell us anything.
A really interesting question is raised here: Glavine’s success? Huh? How did a guy with about-avg strikeouts and about-avg walks post a career ERA+ of 120 (that’s 3.46 compared to a league 4.16)?
He posted a BABIP at the low end of the modern curve (.284), but not too far off the mean (which I’d guess to be about .293). So, he’s given up his share of hits.
The best info I can find at baseball reference is that Glavine does what you’re supposed to do: challenge hitters with no one on base (.254 / .303 / .381), nibble with RISP (.249 / .354 / .347) and especially when there are two outs (.229 / .357 / .316). That .229 avg doesn’t mean he’s been fanning guys in big spots, it means he’s been letting them get themselves out with weak contact (.268 BABIP).
I dunno if simply following the strategic fundamentals of pitching will lower your ERA by 0.70… I’ll look at his flyball and HR rates and get back with more info.
Okay, Glavine’s GB/FB is 59%/41%. I think that qualifies him as a definite GB pitcher (which limits HRs), but not an extreme one.
7.5% of his Fly Balls have resulted in HRs. I think that’s absurdly low, regardless of his 3 home parks (Fulton County Stadium, Turner Field, Shea), but I don’t remember what the avg rate is (11%?).
Stat-head wisdom claims that a pitcher can’t control the % of his flies that become HRs. Maybe Tommy’s slow changeup is an exception to that.
So, I guess we have a hint as to which one of these has been the key to Glavine’s success:
K /9 BB/9 HR/9
5.38 3.04 0.70
If anyone has access to the league rates for those during Glavine’s career, please post.
More possible evidence for change-up masters keeping fly balls in the park, in the form of Jamie Moyer:
tiny Kingdome
yr BABIP HR/FB
1997 .278 9.5%
1998 .286 8.9%
1999 .294 8.0%
spacious Safeco
yr BABIP HR/FB
2000 .303 11.6%
2001 .251 8.4%
2002 .246 8.8%
2003 .269 6.3%
2004 .272 15.1%
2005 .298 7.7%
career .285 10.2
Thanks for those numbers Argon.
bases empty (.254 / .303 / .381)
RISP (.249 / .354 / .347)
two outs (.229 / .357 / .316)
The number that stands out there is the slugging. It’s very low. The league average for .SLG is .427. That stands out to me because of something else I read today about Tommy Gun:
I think that slowing down the bat speed by working that changeup away has a direct impact on the hitters’ ability to drive the ball. Hence the low slugging. Perhaps that’s the stat scout connection that best explains Glavine’s success.
Test