Sabermetrics

[AWvsCBsteeeerike3]

I have never stated, nor will I ever state, that I have a knowledge of the game that even approaches that of a major league manager. That does not however, mean that I don't know more about something specific. I'll give you one example of where I know more than Tony LaRussa.

The notecards. I've heard several comments from TLR himself, read about this in several books (Men at Work, 3 Nights ..) where he basis his decision of lineup construction or PH situations on a handful of AB's by a hitter against a certain pitcher. By handful, I'm saying < 75 or so. I've reread his quotes to make sure I understand the intent, and I think I do. On some occasions, he is basing decisions on the RESULTS of a statistically insignificant number of AB's. Not what he SAW in those AB's, or what a scout SAW in those AB's, but what the RESULTS were.

LaRussa is wrong, and I am right. It's not even debatable. Basing your decision on less than 75 AB's or so is just bad math. Basing your decision on what you SAW in 1 AB - different story entirely.

But that isn't saying that I know a fraction of what TLR knows in his pinky, it's just that I know he is wrong there and I am right.

I agree with you hear, jim. But, unlike "protection", this is not something every manager does or believes in. Not every manager would start Duncan over Ludwick simply because Dunc was 2-5 against somebody while Lud was 0-3.

[Radbird]

If the answer is that it shows that pitchers don't pitch differently in different situations, I can't help but be skeptical about the measurement validity. If it shows that in the aggregate that pitchers tend to pitch the same then what does that tell us? I find it hard to believe that pitchers don't try to vary their approach depending on who is batting and the situation they are in. They certainly claim that they do.

I know I must not be explaining this clearly, because in a couple other recent threads on this topic I get the same sort of counter-arguments as this one.

I am not saying that pitchers do not change their approach. I am also not saying that batters do not change their approach.

I am saying that whether or not they change their intended approach, the overall outcome is pretty much the same. Pitchers very well may change their approach when hitters are unprotected -- and batters may change their approach too, in reaction to the pitcher's changes, or because of their own belief about who is up next. But the end result is not that batters make worse contact and hit less effectively when they are unprotected. Protected hitters do walk a little less often, but they also strike out a little more often, and for balls put in play, protected hitters hit the ball about the same as when they are unprotected. Overall, a hitter's OPS does not change much as a result of "protection." The idea that a protected hitter can HIT better, DRIVE the ball better because they see "better pitches to hit" is not true. When they put the ball in play, they hit it about the same and get about the same results when they are protected as when they are not (or, as I said elsewhere, maybe even a tiny bit worse when protected, as pitchers seek to avoid baserunners ahead of a protected hitter not only by limiting walks, but by limiting hits too -- which in general is the goal almost all of the time anyhow).

I thought you were clear before, but this really sums it up nicely. Thanks.

[clement]

If the answer is that it shows that pitchers don't pitch differently in different situations, I can't help but be skeptical about the measurement validity. If it shows that in the aggregate that pitchers tend to pitch the same then what does that tell us? I find it hard to believe that pitchers don't try to vary their approach depending on who is batting and the situation they are in. They certainly claim that they do.

I know I must not be explaining this clearly, because in a couple other recent threads on this topic I get the same sort of counter-arguments as this one.

I am not saying that pitchers do not change their approach. I am also not saying that batters do not change their approach.

I am saying that whether or not they change their intended approach, the overall outcome is pretty much the same. Pitchers very well may change their approach when hitters are unprotected -- and batters may change their approach too, in reaction to the pitcher's changes, or because of their own belief about who is up next. But the end result is not that batters make worse contact and hit less effectively when they are unprotected. Protected hitters do walk a little less often, but they also strike out a little more often, and for balls put in play, protected hitters hit the ball about the same as when they are unprotected. Overall, a hitter's OPS does not change much as a result of "protection." The idea that a protected hitter can HIT better, DRIVE the ball better because they see "better pitches to hit" is not true. When they put the ball in play, they hit it about the same and get about the same results when they are protected as when they are not (or, as I said elsewhere, maybe even a tiny bit worse when protected, as pitchers seek to avoid baserunners ahead of a protected hitter not only by limiting walks, but by limiting hits too -- which in general is the goal almost all of the time anyhow).

Sorry I hadn't seen any prior threads where you addressed this. But you had brought up the use of pitchf/x and alluded to that as providing input data that could shed light on this subject. But it seems the research you're citing is just about outcomes. However, one of the articles haltz linked provides an input-based statistical basis for a different (opposite) conclusion than the one you are arguing (that protection does exist and has an impact).

http://mlbresearch.blogspot.com/

More generally, the writer of that article makes some of the arguments in a much more eloquent and cogent (and scientific and statistical) way than I do that summarizes my skepticism of many of the (what I feel are) over-reaching conclusions from some people who take a sabremetric approach.

Bill James has noted a troubling historical pattern: sabermetric research often labels a phenomenon (e.g., clutch hitting) as non-existent when it can not be measured or detected. ...

...at bat outcomes involve many moving parts (where the ball lands, reaction of the defense, and luck, to name a few), and Bradbury is trying to measure the effect of an outcome-based rate (OPS) on another outcome. Thus, if there is some noise or randomness within the data, the problem would be compounded in the findings. ...

The sabermetric community will benefit as it moves away from its relatively strict reliance on outcomes and outputs. Events on the field of any sport involve a great deal of processes. While outcome data (e.g., much of what you find online at great sites such as retrosheet and baseball-reference) have generally been more widely available, a full picture of economic analysis in the future will rely much more heavily on whole processes and their inputs.

The author goes on to propose an even more complicated approach to measuring and doing analyses. Personally I think it's mostly an intellectual masturbation exercise and a waste of time, because (a) I already believe a lot of these things (e.g., team chemistry, clutch hitting, protection, leadership) are real, but (b) they aren't so impactful that they warrant this much effort into proving or disproving. But I realize doing this stuff is a lot of fun for many people.

[skmsw]

one of the articles haltz linked provides an input-based statistical basis for a different (opposite) conclusion than the one you are arguing (that protection does exist and has an impact).

OK, so I looked at the article you linked to, and I disagree with your suggestion that it is saying something different than what I said. The article says:

The protection production function seems to tell us conflicting stories. The "input" findings show that protection exists, but the "output" evidence suggests that protection does not exist.

The "input" is whether the batter is seeing more strikes or not -- the "output" is the outcome of the at-bat, whether results are changed with or without protection. This means that pitchers may throw fewer strikes when a batter is unporotected (I said that batters may walk somewhat more), but that the overall results, the actual performance of the hitter, and what happens when the batter puts the ball in play tend to be the same regardless of protection (I said that while batters may walk more when unprotected, when protected they do not hit the ball any better or produce better outcomes). Same thing, no?

Personally I think it's mostly an intellectual exercise and a waste of time, because (a) I already believe a lot of these things (e.g., team chemistry, clutch hitting, protection, leadership) are real . . .

I don't mean to misinterpret you, but are you saying that since you already believe them it is not work reviewing the evidence or determining in a more fact-based way whether you're right or not?

[cards2468]

I have to agree with skmsw here. The article seems to be saying that at bats by "protected" batters are different than when they're unprotected unlike previous research that only used outcomes to say that there's no effect. The article says the only real change is the pitches thrown, but they still lead to essentially the same outcome as if the batter was not protected.

Interesting article though and it does make a good point that conclusions can't be made simply by the output of an event.

[clement]

one of the articles haltz linked provides an input-based statistical basis for a different (opposite) conclusion than the one you are arguing (that protection does exist and has an impact).

OK, so I looked at the article you linked to, and I disagree with your suggestion that it is saying something different than what I said. The article says:

The protection production function seems to tell us conflicting stories. The "input" findings show that protection exists, but the "output" evidence suggests that protection does not exist.

The "input" is whether the batter is seeing more strikes or not -- the "output" is the outcome of the at-bat, whether results are changed with or without protection. This means that pitchers may throw fewer strikes when a batter is unporotected (I said that batters may walk somewhat more), but that the overall results, the actual performance of the hitter, and what happens when the batter puts the ball in play tend to be the same regardless of protection (I said that while batters may walk more when unprotected, when protected they do not hit the ball any better or produce better outcomes). Same thing, no?

Yes, but he also goes on to argue why the output-based approach is "flawed" (his word) and why an input-based approach is more statistically sound and provides better accuracy. He is basically challenging the methodology of the studies that I assume you're citing.

Personally I think it's mostly an intellectual exercise and a waste of time, because (a) I already believe a lot of these things (e.g., team chemistry, clutch hitting, protection, leadership) are real . . .

I don't mean to misinterpret you, but are you saying that since you already believe them it is not work reviewing the evidence or determining in a more fact-based way whether you're right or not?

Well you left off the (b) part of my reasoning, but yeah, for me it's not worth it, because I don't derive any pleasure from it (which would be one reason to do it). But more importantly, I really believe that a lot of these things are not quantifiable or measurable. So it eventually will boil down to what a person believes. X matters or X doesn't matter. Whenever I've gotten into the details of these studies, I see the flaws or limitations in the methodology. I'm not throwing the whole baby out with the bathwater, but I do think that in some instances this has become way too much ado about almost nothing.

Take for example the protection argument. Let's say there was irrefutable proof that protection doesn't affect any batters' outputs in any way. What should a manager or GM do differently? Are you going to put Cesar Izturis batting behind Albert Pujols? Your best hitters bat at the top of the order and your worst ones bat at the bottom.

Taking a page from jim, let's just keep it simple. Get as many great hitters as you can who can play their position and arrange the order with your best hitters getting the most plate appearances. You can tinker with the exact order here and there and it probably won't have much of an impact, either positive or negative.

[jim]

one of the articles haltz linked provides an input-based statistical basis for a different (opposite) conclusion than the one you are arguing (that protection does exist and has an impact).

OK, so I looked at the article you linked to, and I disagree with your suggestion that it is saying something different than what I said. The article says:

The protection production function seems to tell us conflicting stories. The "input" findings show that protection exists, but the "output" evidence suggests that protection does not exist.

The "input" is whether the batter is seeing more strikes or not -- the "output" is the outcome of the at-bat, whether results are changed with or without protection. This means that pitchers may throw fewer strikes when a batter is unporotected (I said that batters may walk somewhat more), but that the overall results, the actual performance of the hitter, and what happens when the batter puts the ball in play tend to be the same regardless of protection (I said that while batters may walk more when unprotected, when protected they do not hit the ball any better or produce better outcomes). Same thing, no?

Yes, but he also goes on to argue why the output-based approach is "flawed" (his word) and why an input-based approach is more statistically sound and provides better accuracy. He is basically challenging the methodology of the studies that I assume you're citing.

Personally I think it's mostly an intellectual exercise and a waste of time, because (a) I already believe a lot of these things (e.g., team chemistry, clutch hitting, protection, leadership) are real . . .

I don't mean to misinterpret you, but are you saying that since you already believe them it is not work reviewing the evidence or determining in a more fact-based way whether you're right or not?

Well you left off the (b) part of my reasoning, but yeah, for me it's not worth it, because I don't derive any pleasure from it (which would be one reason to do it). But more importantly, I really believe that a lot of these things are not quantifiable or measurable. So it eventually will boil down to what a person believes. X matters or X doesn't matter. Whenever I've gotten into the details of these studies, I see the flaws or limitations in the methodology. I'm not throwing the whole baby out with the bathwater, but I do think that in some instances this has become way too much ado about almost nothing.

Take for example the protection argument. Let's say there was irrefutable proof that protection doesn't affect any batters' outputs in any way. What should a manager or GM do differently? Are you going to put Cesar Izturis batting behind Albert Pujols? Your best hitters bat at the top of the order and your worst ones bat at the bottom.

Taking a page from jim, let's just keep it simple. Get as many great hitters as you can who can play their position and arrange the order with your best hitters getting the most plate appearances. You can tinker with the exact order here and there and it probably won't have much of an impact, either positive or negative.

Sometimes I think things are overstated to make a point. Clutch for example ... I'm almost sure that even the SABR community has established that some hitters have this skill. The problem is, it's so small when it does exist that it has no real practical value. So they say "Clutch doesn't exist" to make a point.

I think we all agree that protection changes things, the real question is whether it changes what is important - i.e the ability of a team to score runs. When I say I don't put much stock in protection, I'm not saying that player X won't get walked more or see fewer good pitches etc..., I'm saying that it doesn't really impact the final result.


Here is an opinion of mine, based mostly on what I felt when I played and what I observed in others(several of them future MLB) playing the game. Here goes ...

Baseball is largely an individual game played by selfish players who are in general more interested in their personal performance than the team. This selfishness for the most part is beneficial to the team, and therefore masks itself very well as "team play". But human nature finds this distasteful and invents ways to sugar coat this bitter pill. Batting order and protection are forms of this sugar, creating teamwork from a vacuum.

[AWvsCBsteeeerike3]

Lots of fun stuff in this thread. New blood does a forum good. Now, Steverino's two cents:

I don't actually claim to have "greater knowledge" than anybody. But sometimes I'm right. And sometimes guys like TLR, Joe Torre, and, well, almost every manager in baseball -- are wrong. It's really that simple. They don't always know more about everything, every single time, just because they are on the inside. Some of the people on the inside really shouldn't be.

I think TLR and Torre should be. And, I'm not saying that these guys are never wrong. Just that, well, this flies in the face of what every manager says in baseball, and largely how they build their lineups, I believe. Like I said earlier, I'm not going to simply disregard that in favor of what guys on a mb say simply because sabermetrics - which you all admit have their flaws - says otherwise.

TLR is NOT a stats manager.

See the argument below about notecards. I think he is.

But it didn't seem like the earth could possibly be round, either, once. I don't say that to be demeaning; until we were able to actually sail around it, until we were able to see the earth from a different perspective, it just flew in the face of common sense of even the brightest, most educated people, that it might be round. But sometimes in truth, things simply are not what they seem. Disregarding evidence because it does not fit the preconceived view is a historically poor way to approach new information.

This evidence has been around for decades. What you're saying isn't anything new. It's not like OPS just became available in 2007. Intelligent baseball minds, those that make a living based solely on how their teams perform, have seen the numbers and still believe in protection.

In my humble, honest opinion, statistics serve a few basic purposes. They help us ask better questions. They help us verify OR CHALLENGE what we think we know, what we think we saw. And they help plug the gap that comes from the fact that no one can be everywhere or see everything.

And, like you, I believe they serve their purposes. I think we just differ in what purposes they serve. To me they come secondary to observation mostly. One example is Wainwright. I don't knw off the top of my head, but was there anything to suggest that he was going to be such a bad ass when he was coming through the farm system? I don't really think there was (other than the low minors) and he was projected as a #3 or at best a #2 starter. And, that's how he came out in 2007. But, just watching him pitch, you had to know he was going to be a beast. Would sabermetrics ever have projected him as an ace? I honestly don't know? But just looking at his stuff, it was highly likely he was going to be a top of the rotation starter. throwing in his mental makeup - which you can never get from stats - it was, in my mind at least, guaranteed he was going to be a bad ass.

Don't be too sure. Some owners, GMs, and managers have very sophisticated, mathematically validated, carefully researched systems -- and others would still, to this day, make Grady Little look like a Rhodes Scholar. But don't assume that just because someone makes a living in baseball and uses "concise" stats, that those stats are valid.

Sort of confused....and don't want to say anything stupid. So, I'll just say good point.

Why are you so bent on making this either/or? For all the flak that "stats people" get for "not believing in scouting" or"not watching games," I pretty much only see these either/or positions come from people who devalue objective measures of performance and data.

Conversely why are you so against observation. When TLR comments that AP has fewer homeruns because he's getting pitched differently, I actually believe that because to me, at least in the beginning of the year, it made sense.

Why are you viewing "sabermetrics" as something separate from 100+ years of baseball knowledge, rather than a continuation and advancement of it?

Because, in this case at least, if they were the same, we wouldn't really be arguing about it.

So, most of the time, what you believe and what the numbers say match up; in those instances the numbers are fine. But the time the numbers do not tell you what you expect, you figure that time, the numbers are wrong and you're right?

For 75 years, baseball did not let people of color play; were the people using "logic from the beginning of the game" right about that? Or, was more forward kind of thinking necessary?

Not letting people of color play was a societal issue, not baseball strategy. There is a difference there. And, like I've said before, sabermetrics have their place. However, they're not infallible. This is where you and I differ. You believe, that the stats are the end all be all of baseball. And, I don't.

Lots of reasons maybe -- random variation, for one. Regression to the mean, for another. Steverino's Axiom of Hall of Fame Caliber Players, for a third.

It's taken me long enough just to address your other points.

Don't have time to read that right now, although I will.

a) If his "approach has changed," is that not under his control? Tell him to change it back!!!
b) Who says HRs are completely independent of everything?
c) Hitting less fly balls is not a throwaway point. Getting the ball into the air instead of on the ground is dependent on many things -- how hard you hit it, how much topspin you give it, your batspeed, your reaction time, how well you center the ball on the bat, how well you recognize the pitch, and of course, random variation (a 300 hitter does not get three hits every ten at bats) and chance events (a lot of luck comes into play when you're using a round bat to hit a round ball coming at you at incredible speeds).

a) This is my personal opinion that doesn't really fit into our argument. So, that can be saved for another day.
b) You all did when you said there is no such thing as protection and the types of pitches/way a pitcher pitches to a batter/way a batter approaches an AB is independent of who is batting behind the batter in question.
c) So, if his HR/FB is the same as always and he has less HR's then we can say he is hitting less flyballs, correct? If that is the case, don't you see that as not good?

*Breathes in* *Breathes out* Whew. Good points and arguments, Steve!

[Leroy]

When I was in tee ball, the other team's coach would really set the ball on the tee nicely for me, as I had a really good hitter batting behind me.

[skmsw]

but he also goes on to argue why the output-based approach is "flawed" (his word) and why an input-based approach is more statistically sound and provides better accuracy. He is basically challenging the methodology of the studies that I assume you're citing.

Except he got the exact same results.

And if there is agreement that in the end the results were not affected, I'm not sure I buy his attempts at challenging the methodology. Some studies that have been done, Bradbury (which this guy cited) in particular, control for an awful lot of variables.

I really believe that a lot of these things are not quantifiable or measurable.

Based on my interest in understanding how different things contribute to outcomes, I still keep trying. And watching others who try. I happen to think some pretty good stuff has been done. There is certainly some stuff that can't presently be measured or isn't presently understood; but the academic in me does not let me respond to that by deciding to just stick with what we already think.

Meanwhile, it may be that some of those things are not quantifiable because they have no effect on outcomes.

Let's say there was irrefutable proof that protection doesn't affect any batters' outputs in any way. What should a manager or GM do differently?

I'm not sure. Maybe nothing, if the goal is always to get as many quality players as possible anyway.

but let me turn your question around: suppose, as has been suggested elsewhere in this thread, that Albert is intentionally making changes to his approach out of a belief in protection (or lack of it). In this scenario he is swinging at worse pitches than usual because he thinks that since no one is behind him, he will hit differently, and in a self-fullfilling prophecy, since he is swinging at worse pitches, he is hitting worse, and since he is swinging at worse pitches, pitchers keep giving him slop -- not because of who is up next, but because he is chasing it so they keep doing it. I'm not saying any of this is happening, just suppose. And I'm not saying he has become a .250 hitter; just doing worse than he can, by not being as selective, not waiting for the pitch he can drive the way he is capable of because he does not believe it is coming.

Wouldn't your hitting coach's next conversation with Albert be different based on whether Albert's "adjustments" were a valid response to lack of protection, verses a misguided strategy in response to a problem that does not exist? "Hey Albert -- take the walk, when they see you won't chase they'll start coming closer" verses "attaboy for expanding, Albert, they won't come to you, you go to them! Albert swinging at worse pitches thus doing less with them is still better than albert walking and the next guy coming up with Albert on first, albert won't get any pitches to drive anyhow, swing if it's close and put up with not waiting for one you can wallop" ?

Taking a page from jim, let's just keep it simple. Get as many great hitters as you can who can play their position and arrange the order with your best hitters getting the most plate appearances. You can tinker with the exact order here and there and it probably won't have much of an impact, either positive or negative.

On this I agree with you both.