QOC: The Next Great Metric (Or Should Be, Anyway)
by Nico on Sep 4, 2010 7:00 AM PDT
Email
PrintA disclaimer: If you were at Chez Nico: Lasso Your Own Dead Animal Yee Haw! and it seems like I'm stealing an idea from emperor nobody, just consider that perhaps the ideas were flying around so fast and furious that maybe you lost track.
As many of you know, I have some skepticism about the validity -- or should I say "dynamicness" -- (no, having typed it out I definitely shouldn't make up that word) of the batted ball profiles we currently have available to evaluate and predict. It's not a matter of being "anti stat" but rather a matter of feeling like the batted ball profiles have not yet reached the levels they need to reach in order to sufficiently explain what we are seeing.
At Chez Nico, someone -- maybe it was emperor nobody, maybe it was me, really who can keep track? Ok fine, it was emperor nobody -- used a term that described exactly what I feel is needed to explain why what pitchers like Trevor Cahill and Dallas Braden are doing is not only a product of luck, defense, and park factors. They are also a product of lower QOCs.
QOC stands for "Quality of contact" and if we had a better way to measure it, more precisely, than "line drives and not line drives," "grounders and fly balls," etc., we might better understand why some pitchers thrive more or less than others do, more or less than they "should." Follow me for some FAQs about QOC, i.e., questions that are frequently asked about something no one has ever discussed. OK, so they're not FAQs, they're QTWPBFA: Questions That Would Probably Be Frequently Asked...
"How is QOC measured?"
While this is arguably the most important question, it's the one that won't be answered here today. You can imagine any method to distinguish harder contact from weaker contact, be it a 1-5 rating scale, trajectory and/or velocity tracking, etc. All that matters for the purpose of this discussion is that quality of contact is measured, in whatever way we deem most valid, in order to produce a "QOC" metric.
"How can a pitcher control QOC?"
I'm guessing that Cahill has not picked Craig Breslow's brain about the arc, torque, and spin that will, using advanced Physics and Calculus, engineer a pitch that is hit on the top of the ball in such a way as to cause weak contact on the ground.
Rather, some pitchers may, by good fortune, have through their mechanics late movement or deception that happens to reliably induce generally weaker contact. Cahill might be a shining example of this. Meanwhile, other pitchers may more intentionally induce weak contact by effectively changing speeds and getting hitters out on the front foot, or by effectively using the inside corner to jam hitters. Braden's changeup, and cutter, are excellent examples of this in action.
"Why are line drive rates still similar, and with large fluctuation, for pitchers with low QOC?"
My sense about line drives (and I'm not claiming any certainty here) is that batters influence a portion of this and that the portion influenced by pitchers is largely "mistakes" -- in other words, Cahill isn't giving up a ton of line drives on his "low QOC-inducing power sinkers," but rather he's giving up a lot of his line drives on hanging changeups, spinning curves, and flat sinkers.
So what we're seeing in line drive rates is as much the normal presence and fluctuation of mistakes (and just good hitting) as it is related to the forces that are determining high or low QOCs in general.
"Why wouldn't low QOC pitchers simply have low BABIPs, high QOC pitchers high BABIPs?"
There isn't a perfect correlation. A low quality of contact means fewer grounders will get through to the outfield and fewer fly balls will get through for extra bases, but will also yield more infield hits and bloop hits. Conversely, a high quality of contact means more ground ball hits and extra base hits yet fewer infield hits and bloop hits.
However, more routine plays and fewer balls that get through the infield or outfield are, in the balance, a lot more gooder than the increased dribblers and bloops are bad. So while it won't always show up dramatically in BABIP, or in "line drive rates," your "low QOC" pitchers should ultimately enjoy more success for their ability to induce weaker contact -- whether it's because of how their stuff happens to move or how they strategically work hitters.
Do you believe that different pitchers have potentially measurably different QOCs? And if so, do you believe there is a correlation between lower QOCs and better results?
36 comments
|
1 recs |
Do you like this story?
Comments
I have no idea if this is a good idea or if it's already been studied and dismissed
But you’ve presented the idea well.
Trying to quantify this sort of thing is exactly what sabermetricians do all the time.
www.zekeishungry.com
by thejd44 on Sep 4, 2010 7:14 AM PDT reply actions
Thank you -- that's unusual for me!
I like Cindi. A. She never pretends to know more than she does. B. She has unbridled enthusiasm for her "Hotties," and isn't afraid to show it. -IM4Oakgal
I remember some hardball times thing?
BABIP is consistently lower with two strikes, as if batters had to behave differently when protecting against the strikeout.
here's one
http://www.hardballtimes.com/main/fantasy/article/how-much-do-counts-affect-babip/
and the answer from that article is that BABIP varies by count, but not enough to matter for actual counts by actual pitchers.
This is one reason why I choose to not put much stock in some sort of assumed fixed BABiP...
…nor really attempt to understand why it should be assumed fixed in it’s application. As I’ve posted before, and for the same reasons you’ve stated here, I think the assuming that BABiP stays relatively fixed for all pitchers just misses too much.
Great thought provoking post. Rec’d!
Interesting and thought provoking.
Hit f/x should help with this as well.
A's Fan in Sweden
"Sosa had me caught up in the magic, and I feel like an idiot. I don’t say that often, but I feel like an idiot because of Sammy Sosa." -Jay Mariotti
I was gonna post hit f/x as well
When we have data on the exact speed and trajectory of balls off the bat, we’ll have a much more reliable data set.
Foul percent might correlate with QOC numbers?
Of course in Oakland you would have to think about ballpark details…
Last night for example most of the foul outs
Would have been foul outs in any ballpark. I don’t know what % of pop fouls at the coliseum would be in the stands elsewhere but I suspect it’s not as large as people might think.
by OaklandSi on Sep 4, 2010 9:33 AM PDT via mobile up reply actions
If it's easier to foul out then paradoxically you might see a lower foul%
for Oakland pitchers who would otherwise have excellent QOC numbers. My guess is that pitchers like Braden might have a higher number of foul outs than a typical pitcher…
The pitchers themselves distinguish
Between those who “miss bats” and those who “pitch to contact”. Braden and Cahill are examples of the latter. Obviously if you’re pitching to contact you want poor contact.
by OaklandSi on Sep 4, 2010 9:30 AM PDT via mobile reply actions
I can't remember either
We were all in the hot tub and there was a lot of drinking involved
I think it was “Quality of Cognac”
I'm here to talk about Don
It was actually chlorinated water
Oh, and my hot tub is only half full now, jerk.
I like Cindi. A. She never pretends to know more than she does. B. She has unbridled enthusiasm for her "Hotties," and isn't afraid to show it. -IM4Oakgal
This is a great idea, but it's certainly not a new one.
This is basically exactly what everybody’s going to do the minute HitFX comes out. That system is supposed to give us precise measurements about the launch angle and launch speed of every batted ball that’s hit. There’s a bunch of really cool possible applications, and pitcher quality of contact is certainly one of them. I’d imagine you could build a pitcher stat that’s based entirely and only on HitFX launch data using “buckets” much like UZR, which would be the purest example of a “fielding independent” stat I can think of.
Like topspin off of the bat?
I’m not sure about that one. Don’t know if the cameras have time to pick up spin. With PitchFX, they’ve got a whole flight path to look at. With HitFX, they’re just looking at the area directly around the batter.
just based on the physics of the bat-ball collision
any time you are hitting the ball with a lot of spin, you aren’t making solid contact. Of course, even a little bit of spin can affect things (i.e. hit the ball hard, but with a bit of back spin, to get more loft on a HR ball).
You know when you've hit the ball on the screws...
…when you send a screaming ball knuckling into the outfield on a line. I’ve had to try and catch a few of those as a CF and it isn’t all that easy.
by LowcountryJoe on Sep 5, 2010 9:27 AM PDT up reply actions
does uzr or dewan consider distance traveled to the ball by the fielder?
"The ego, the super-ego, and the Ed" - dannycakes
Actual distance? No, but they have a funky system that sorta works.
They separate all batters into “powerful” or “not powerful”, and assume outfielder depth based on that. For infielders, they do a similar thing, but based on the handedness of the batter, the speed of the batter, and any shift present.
OK, so, we can't currently measure QOC very well
yet, somehow we know that Cahill and Braden have it.
Talk about assuming the conclusion.
This is the Carlos Gonzalez saga all over again. To recap: someone posited that sucking at hitting curveballs might in theory be able to generate an unusually large park effect for him. That got transfigured into “Carlos Gonzalez sucks at hitting curveballs and would be awful in Oakland,” because that’s what people wanted to read out of it, despite the utter lack of evidence for any such conclusion.
Elcroata then demonstrated that that hypothesis was, in fact, utter nonsense, and all the variation from the norm was on fastballs (and very carefully and correctly did NOT assume that that variation was the result of something specific about Coors Field). Naturally, the reaction was “Carlos Gonzalez sucks at hitting fastballs and would be awful in Oakland.”
Let’s review certain facts here.
1. Ground balls are not subject to “QOC” classification problems to nearly the degree that line drives and fly balls are. A ground ball is a ball that bounces. Assuming the stringer can see, he’s got that one down pat.
2. The effects of ground balls are (correctly) evaluated into the tRA and xFIP metrics, which, as I’ll remind people, correlate much better than ERA does to the next year’s ERA.
3. While pitchers have different BABIP skills within a fairly narrow range, it takes an enormous number of plate appearances before one’s estimate of BABIP becomes confident. Adding six years’ worth of innings at a league average .295 BABIP spits out an expected figure, for Trevor Cahill, of about .287. That’s barely noticeable— even over a full season, it’s a rounding error. He’s even closer to league-average when you regress toward a number which is informed by the park and defense of the A’s, which is probably more like the team average of .280.
4. While it may be that Cahill (I’ll ignore Braden because his career BABIP is, if anything, higher than one would expect) is actually producing balls with lower velocity (or topspin, or whatever) off the bat, there’s literally zero evidence that he is. Low BABIP is not evidence of low QOC.
It’s one of these correlation-causation things, but it’s amplified by the fact that pitchers have lucky seasons all the time— and then promptly revert back into their previous non-awesome selves. Last year’s BABIP leaderboard included such luminaries as Ross Ohlendorf, Randy Wolf, and Jarrod Washburn. 2008? How about Dave Bush, Armando Galarraga and Greg Smith?
5. Pitchers who are very much consciously aware of BABIP as a phenomenon have attempted to make changes to their pitching style to improve it, and to my knowledge, all of them have failed miserably (the most prominent example being fringe arm and occasional sabermetric folk hero Brian Bannister).
Bottom line: Yes, some pitchers seem to be able to generate lower QOC than others; there is zero evidence that Cahill is one of them, and given our technology, there is no way there could be at this point in his career.
Technological improvements may help with this (perhaps cutting down the amount of regression to the mean to maybe 2-4 seasons’ worth of averageness instead of 6 seasons), but they are not going to be a panacea, because the two basic facts about BABIP (Fact one: round bat + round ball = funny bounces; Fact two: pitchers cannot control who steps into the batter’s box against them) will continue to exist as long as the game of baseball does.
"We don't want our people to be preoccupied with seminude, crazy men jumping up and down who are chasing an inflated object," said Sheik Mohamed Osman Arus, head of operations for the Hizbul Islam insurgent group.
Some good points here, but it's not a C. Gonzalez argument all over again
1. I don’t agree. All ground balls are NOT alike. To give examples in the extreme, a pitcher whose average 10 ground balls are 6 sharp, 2 routine, 2 dribblers will have a very different experience from one whose average 10 ground balls are 2 sharp, 6 routine, 2 dribblers. A pitcher may not know how to intentionally get different QOC ground balls, but his pitching may well induce different average QOCs from the pitcher next to him.
2. I don’t know of anyone who is arguing that ERA is a very good predictor of ERA. Metrics like FIP and tERA predict ERA far better than ERA does. Is this in question? And perhaps a QOC component could help the next metric predict even better than FIP or tERA does. Si?
3. I’m actually suggesting that BABIP isn’t a big part of the QOC conversation. I think I’ve expanded on that in the post.
4. Correct: There’s currently zero “evidence” that Cahill is producing more weak contact than average. That’s because we don’t have a metric like QOC to provide that evidence. If we did, we’d now have evidence to better understand Cahill’s pitching profile.
5. Like ERA, BABIP is a bad tool to use to explain or change what you’re doing.
Bottom line: Your “bottom line” is an endorsement of the development of QOC, so that we can see who are the pitchers generating lower QOC and then can determine whether a given pitcher like Cahill is one of them and have more understanding of what is, or isn’t, driving his successes and failures. Thank you for your support! ;-)
I like Cindi. A. She never pretends to know more than she does. B. She has unbridled enthusiasm for her "Hotties," and isn't afraid to show it. -IM4Oakgal
1. I certainly endorse the development of a QOC metric once one becomes available
and I think the point that the LD/FB distinction is awkward is well taken, which is one reason that I put very little stock in players’ nominal LD rates when I’m analyzing their play.
I’m less optimistic about getting something out of the velocity of ground balls once we know that. I suspect that the distinction between high and low-velocity grounders is analogous, for pitchers, to the distinction between K’s and in-play outs for hitters. High-velocity grounders will go for hits more often but they will also produce more double plays.
All of which is of course speculative until we see the data.
2. Yes, improving on the current QOC metric in tRA would improve its predictive ability.
3. Offense is four things: Ks, BB/HBPs, slugging, and BABIP. Two of those are directly implicated in the various fielding-independent metrics. Slugging is very heavily implicated through HR rates, though there are some minor aspects of it around the fringes that are not grasped. The whole point of this business is to try to predict BABIP better, because that’s the area of pitching that is most opaque to us— though there may be some incidental benefits to our ability to predict slugging as well, I suppose.
4. If all you’re arguing is “QOC-metric-good”, then Trevor Cahill should not be appearing in this post. Insinuation-by-juxtaposition might be a nice rhetorical trick but last time I checked, the point of all this blog business was not to trick people into believing you. I don’t object to saying “we need more evidence,” I object to saying “we need more evidence so that we can prove that x is true” when, in fact, you have no idea whether x is true or not.
Teleology produces bad analysis. I think that’s very obvious at this point (Salem witch trials; McCarthyism; Duke lacrosse case; the list goes on).
"We don't want our people to be preoccupied with seminude, crazy men jumping up and down who are chasing an inflated object," said Sheik Mohamed Osman Arus, head of operations for the Hizbul Islam insurgent group.
Just add "or isn't" --
"we need more evidence so that we can prove that x is true, or isn’t." I’m not using any rhetorical tricks; I’m just using pitchers that are familiar to A’s fans as examples of pitchers who have had success that different metrics (if we had them) might better explain.
Though a QOC metric wouldn’t prove that Cahill’s low QOC (if he had one) explained some, or how much, of his success. That requires extrapolation: “Pitchers with lower QOCs are, on average, better than pitchers with higher QOCs…Cahill has a low QOC…Therefore, Cahill’s low QOC explains his success…”
I like Cindi. A. She never pretends to know more than she does. B. She has unbridled enthusiasm for her "Hotties," and isn't afraid to show it. -IM4Oakgal
Well, that's fine
Just say that then. Saying “x is true, or it isn’t” is very, very different from saying “x is true.”
On an unrelated note, it would not shock me if Cahill had actually asked Breslow the mock question in your post. The A’s surely have to have the most intellectual/ eggheaded group of pitchers in baseball— Cahill was committed to Dartmouth before signing with the A’s, Jerry Blevins expects to go to law school after his baseball career wraps up, Andrew Bailey went to a decently selective liberal arts school, and Brett Anderson is the son of a pitching coach. In the minors, there’s Shawn Haviland, who went to Harvard.
It’s enough to make me wonder if there’s a deliberate strategy behind it.
"We don't want our people to be preoccupied with seminude, crazy men jumping up and down who are chasing an inflated object," said Sheik Mohamed Osman Arus, head of operations for the Hizbul Islam insurgent group.
"QOC could help explain why Cahill is getting good results"
is a sentence I can write or read. “QOC could explain why some pitcher — I won’t name any right now because that would be irresponsible — might be (or might not be, we don’t know yet because we don’t have the metric in question yet) pitching well” is a sentence I’d prefer not to write or read.
As for intellect, Ziggy and Braden are also exceptionally articulate and reflective. But son of a pitching coach or not, Brett Anderson doesn’t exactly come across that way to me.
I like Cindi. A. She never pretends to know more than she does. B. She has unbridled enthusiasm for her "Hotties," and isn't afraid to show it. -IM4Oakgal
Isn't teleology everywhere?
…I object to saying "we need more evidence so that we can prove that x is true" when, in fact, you have no idea whether x is true or not.
Teleology produces bad analysis…
Here’s the thing, though, almost no one attempts to prove something that goes against their own beliefs and biases. To a certain extent, many more people than not come up with a plausible correlation and then spend their time trying to show statistical significance. If it works out, it is shared and published. If it doesn’t, some keep on tweaking the model or they give up. It’s very rare that people share/publish their findings when their models cannot show causality; and it get’s increasingly more rare the deeper the belief and biasies are held.
I think it is just as bad to not share or publish research. I often wonder how much time is wasted because of this problem.
If all you’re arguing is "QOC-metric-good", then Trevor Cahill should not be appearing in this post. Insinuation-by-juxtaposition might be a nice rhetorical trick but last time I checked, the point of all this blog business was not to trick people into believing you.
I think TC is a pretty good example to use to suggest that perhaps the metric being used has some limitations. And that having some additional explanatory variable(s) may be helpful. That’s all I read into it. Why do you view it as some kind of a trick?
by LowcountryJoe on Sep 6, 2010 6:37 AM PDT up reply actions
Interesting topic, Nico.
I remember reading several interviews with A’s officials over at the scout.com A’s site (oaklancdlcubhouse.com) where the officials indicate that the organization definitely keeps track of the quality of contact that hitters in the organization make. I would assume that they do the same thing with pitchers.
The only real prominent example of how this stat may have been used recently to “uncover some undervalued asset” is the case of Cliff Pennington…judging by his 2009 minor league numbers, it certainly seemed like Cliff was not ready to hit at the major league level (.264/.345/.367) with a relatively normal .300 BABIP. However, he was promoted anyways for the last two months of the season and put up better numbers at the MLB level than he did in AAA and is holding his own this season as well (99 wRC+).
Perhaps the A’s saw that his quality of contact in AAA was better than his numbers – even BABIP – would normally indicate and decided that he was ready.
Maybe this a poor example, but it seems like it would be a useful stat that could be used all the time.
I'm never gonna do it without the fez on!
Thanks; I agree.
Interesting to know that the A’s do track this in some way.
I like Cindi. A. She never pretends to know more than she does. B. She has unbridled enthusiasm for her "Hotties," and isn't afraid to show it. -IM4Oakgal
Yeah, this is what people are waiting on Hit-F/x for
It’s really for hitters, not pitchers, where the metric promises to make major improvements in evaluation. (Though it’s unlikely to come to the minors soon, so prospectors will continue to have to rely on proxies and scouting reports.) Unlike pitchers, hitters DO have widely differing talents at BABIP (compare Ichiro to Carlos Quentin) and it would be nice to start figuring out why, and whether sudden spikes and canyons in BABIP are the consequence of something fundamental about someone’s hitting, or of random swings of fortune.
"We don't want our people to be preoccupied with seminude, crazy men jumping up and down who are chasing an inflated object," said Sheik Mohamed Osman Arus, head of operations for the Hizbul Islam insurgent group.
This could be true regarding Pennington, but I wonder if it's mostly just health
He’s been remarkably durable in the majors, and was mostly the opposite in the minors.
www.zekeishungry.com
by thejd44 on Sep 5, 2010 9:26 AM PDT up reply actions
oh, man
I meant it like, when they are smoking weed in the spot and one blows the hit in the other’s face… you know, how high one gets from the other’s exhaled smoke, Quality of Contact.
A true sabermetrician would break this stat down in terms of organic outdoor vs. organic indoor vs. hydro herb.
Silence s'il vous plait!! Vous ne voyez pas que je suis en train de se masturber?!?
You're thinking of "Quality Of Cannabis"
There you want a high number.
I like Cindi. A. She never pretends to know more than she does. B. She has unbridled enthusiasm for her "Hotties," and isn't afraid to show it. -IM4Oakgal
It's completely subjective with qualitative measures
Now if we’re talking hit rates, then we’re on to something.
by LowcountryJoe on Sep 5, 2010 9:30 AM PDT up reply actions
Wow -- this schtick was a great joint effort.
I like Cindi. A. She never pretends to know more than she does. B. She has unbridled enthusiasm for her "Hotties," and isn't afraid to show it. -IM4Oakgal
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
