Movement Rating

[Luis_Rivera]

The latest DIPS study (found in THT's 2006 baseball annual) contained a ****load of developments. One of them really highlights a problem that has been touched upon with the pitcher ratings in this game: Movement being what effects HR.

It is already known that HR are affected by how many fly balls a pitcher gives up. However, that particular study was able to go a little deeper. While the year to year correlation coefficient for both hitters and pitchers when it comes to fly ball percentage is above .7 which indicates a strong relationship (indicates that both control how many fly balls are hit), only hitters have a relatively strong (I can't remember the exact # and my book isn't with me at the moment) year to year relationship with regards to HR/FB.

The pitchers' correlation coefficient for year to year HR/FB was a measly .08... This indicates that it depends almost entirely on the hitter and the stadium how many of those balls are going to make it out of the park.

So, my point is that movement should just basically be removed from the equation. I understand that this is quite a drastic change, and implementation would probably be quite a process, but I figured I'd get the ball rolling given the developments over the last few months in DIPS, which have been many.

I'll post all the related corellation coefficients to this issue either when I get home tomorrow nite or Saturday, depending when I'm going out.

[andymac]

So, basically, we are finding out that a pitcher has control over even less than what we thought before? Basically only strikeouts, walks, & whether the ball is hit in the air or on the ground?

[Luis_Rivera]

Originally Posted by andymac

So, basically, we are finding out that a pitcher has control over even less than what we thought before? Basically only strikeouts, walks, & whether the ball is hit in the air or on the ground?

I'll post all the correlation coefficients that are there when I get home tomorrow, I can't remember if there were any developments in the opposite direction.

[andymac]

Also...a little off-topic, but I've got a question about the THT annual.

It says that it "has an in-depth look at the minors". I'm just wondering what kind of things it is looking at. I'll probably purchase it either way, but what content is in there will probably affect how quickly I get around to it.

[Luis_Rivera]

Originally Posted by andymac

Also...a little off-topic, but I've got a question about the THT annual.

It says that it "has an in-depth look at the minors". I'm just wondering what kind of things it is looking at. I'll probably purchase it either way, but what content is in there will probably affect how quickly I get around to it.

I'm trying to remember what I read in it regarding the minor leagues, but I honestly read so much **** about baseball that I can't remember where I read what. I'll shoot you a PM either tomorrow or Saturday when I'm home with what you're looking for.

[RonCo]

Ultimately, since we're not getting ratings for each pitch type), it seems that the eventual best scenadio would be if a pitcher had 5 key ratings:

Strike Outs
Walks
Fly Ball %
Line Drive %
Ground Ball %

I haven't read in detail the latest deep thoughts on pitchers' ability to control line drives, but I'm thinking that some influence has been proven ... though tiny. Perhaps this is a "qualifying" rating--a rating that starts low (or negative) in the minors and eventually gets to the level where he can survive in the majors and the rating levels out for all but a tiny few (for who it continues to improve).

This would be preferable to Stuff, Control, and Movement because the ratings descriptors actually define something the pitcher controls.

An alternative is to give us the weakish scouting nomenclature (Stuff, Control, and Movement), but give us detailed stats on gb/fg/ld so we can figure these "hidden ratings" out. Note--that would mean gb/fb/ld actually did mean something within the game, which I'm assuming they will now.

[Malleus Dei]

Originally Posted by RonCo

Strike Outs
Walks
Fly Ball %
Line Drive %
Ground Ball %

IMO this would be pretty much ideal.

[Luis_Rivera]

Ok, here are the related corellation coefficients for pitchers/hitters (typically .5 indicates a significant relationship, .7 indicates a strong relationship):

Pitchers/Hitters-
K rate: .77/.84
BB rate: .69/.81
HbP rate: .38/.71
HR Rate: .28/.76

So there is a strong amount of control by hitters in all 4 of these categories. Pitchers also have a lot of control for BB and K rates. HbP and HR rate though, look like there is decent amount of control by pitchers. When it comes to HR, though, what control is displayed by pitchers probably comes from FB%.

Pitchers/Hitters-
LD%: -.03/.10
GB%: .79/.73
Total FB%: .73/.73
OF FB%: .69/.71
IF FB%: .50/.60

Looks like LD% is completely random for pitchers (so the line drive % rating for pitchers really wouldn't be much of a very good idea) and while there isn't all that much, there is some by hitters. This is probably a good part of why BABIP fluctuates so much as 75% of line drives go for hits.

Pitchers/Hitters (outcomes of OF flies)-
Outs: .05/.60
Singles: -.08/.03
Doubles+Triples: .08/.22
HR: .08/.77

So, there ya go... HR are almost completely controlled by hitters when you factor out FB%. Here's a little unexpected tidbit, though... it seems that hitters do actually have a significant amount of control over how many of their fly balls (and as you'll see later ground balls) go for hits, as is shown by the high correlation with outs.

Pitchers, though... it seems have almost zero influence on the outcomes of fly balls.

Pitchers/Hitters (outcomes of line drives)-
Outs: .08/.14
Singles: .15/11
Doubles+Triples: .05/.14
HR: .04/.37

As expected, again.. pitchers don't show much, if any, control over outcomes. I think the higher numbers here, especially singles come from the fact that many of the pitchers from the study played on the same teams from year to year and the defenses behind them actually control how many of these balls go for hits more than the pitchers.

Pitchers/Hitters (outcomes of grounders)-
Outs: .17/.46
Singles: .12/.16
Doubles+Triples: .02/.14

It seems, again that hitters have a decent degree of control as to how many of their grounders go for hits. I would imagine that's because the biggest factor is how hard they hit the ball (and how fast they run would get them some infield hits a la Ichiro).

Pitchers are showing even more control here, but again... I'd say that's mostly related to defense. Even moreso than liners, defense (and the playing surface of the home stadium for that matter) play a huge role in how many balls go for hits.

There's also percentages of occurance for each event. I'll post those here when it becomes relevant.

[RonCo]

This is great stuff, LR...where did the data come from?

[Luis_Rivera]

Originally Posted by RonCo

This is great stuff, LR...where did the data come from?

The Hardball Times' 2006 baseball Annual. This article alone was worth the $15 or whatever I paid for it.

[Luis_Rivera]

Originally Posted by andymac

So, basically, we are finding out that a pitcher has control over even less than what we thought before? Basically only strikeouts, walks, & whether the ball is hit in the air or on the ground?

Here's a little bit... the significance of which escaped me the first time around... If you look at the correlation coefficient for popups for pitchers and hitters... they're .5 and .6 respectively.

I put 2 and 2 together before for the hitters and realized how this affects BABIP as 98% of popups are caught for outs (contrast that with 25% of line drives). However, I'm now realizing that this is really the one known way that pitchers can and do control BABIP as there is a significant relationship from year to year with a pitchers' ability to induce popups.

Voros McCracken's initial findings were that knuckleballers tended to have a lower BABIP. Earlier this year I also remember reading a study on Baseball Prospectus that talked about how some guys who have a good change up tended to have a lower BABIP.

Looking at Tim Wakefield (career BABIP of .283), his popup per fly ball numbers are consistently above average. While Mariano Rivera doesn't have an outstanding changeup, he's a guy that controls his BABIP by inducing many popups as well.

Just thought I'd toss this out there given the continual trend finding evidence that pitchers don't control much. This is a finding in the opposite direction. I'd actually like to see a study done between popup% and line drive% (for both hitters and pitchers). I wonder if there's a negative relationship between the two. I think I'm going to shoot an email to David Gassko regarding this.

Edit: nevermind that last part about the relationship between LD% and Popup%... I just took a cursory glance at some of the LD% numbers versus IFFB% numbers and there really appears to be zero relationship between the two. It was a nice thought, though!

[mrbill]

Hmm, I think about my old FPS BBPro days and their GF rating for batters and pitchers.

It wasn't a fixed sort of split, in fact, the GF rating was more representative of the average direction the ball travelled off the bat.

At first glance, it seems like the same thing as OOTP's GB% rating (from what I see of it), but in the BBPro sense, it represents a much more concrete thing that describes all types of hits and their likelihood based on a particular hitter/pitcher matchup.

Basically, nothing in OOTP now distinguishes the angle of a fly ball, just that X% are going to be fly balls in some direction, but the difference between liner and pop-up is significant (as shown in the above posts).

[Luis_Rivera]

I feel like, though... in order to have an adequate DIPS engine at this point you could still make BABIP essentially random for pitchers and use Contact as a BABIP rating to go hand in hand with Avoid K's as I suggested in this thread. When it comes to pitchers at this point, my biggest issue is the FB% and the movement rating.

I really don't care about the components of BABIP at this point with OOTP... Everything can be implemented one step at a time (starting with a few very broad ratings) if that's easier, for all I care. I'm patient enough. I'd just like to see some steps toward the most realistic model possible.

[Skipaway]

These kind of correlation actually is just a representation of the balance among pitching, hitting, and defense in the current MLB environment. The correlations would be vastly different if we change the elements.

For example, the variety of pitches in the league could change the relationships easily.

Also, "findings" like pitchers having little contrl on fly balls actually just means how good the defense is at fielding fly balls.

[palli]

Originally Posted by Skipaway

Also, "findings" like pitchers having little contrl on fly balls actually just means how good the defense is at fielding fly balls.

That should be not right since FB are all balls hit into the air excluding only LD but it does not matter if the ball is caught by a defender or lands for a hit (or HR). So defense has absolutly no (ZERO) influence on the FB% of a pitcher.

palli

[Skipaway]

Originally Posted by palli

That should be not right since FB are all balls hit into the air excluding only LD but it does not matter if the ball is caught by a defender or lands for a hit (or HR). So defense has absolutly no (ZERO) influence on the FB% of a pitcher.

palli

I am not talking about the fly ball percentage. I am talking about if the fly balls would be turned into outs.

[palli]

Originally Posted by Skipaway

I am not talking about the fly ball percentage. I am talking about if the fly balls would be turned into outs.

Sorry, I did not understand your post correctly.

But the article has it as you write: Pitcher have absolutly no control if the FB falls for a hit or not. It is all defense and to a smaller amount the batter.

palli

[Skipaway]

Originally Posted by palli

Sorry, I did not understand your post correctly.

But the article has it as you write: Pitcher have absolutly no control if the FB falls for a hit or not. It is all defense and to a smaller amount the batter.

palli

And I am saying, if we got superman defenders, pitcher would have absolutely no control if other kind of balls would be a hit or not either. Right now there are obvious differences, because the defense couldn't handle groundballs or linedrives as well as flyballs.

The most significant factor would be flight time, which would limit the fielder reaction time. I would say the better the defender, the less difference between linedrives and flyballs. Groundballs varies more in speed, but they'd definitely be slower than linedrives.

Now if we consider what makes a ball linedrive or flyball, we'd see it's mostly about the spin at contact and contact angle. That's why the kind of pitch a pitcher is throwing might be key. I doubt the result would be the same in a league with mostly fastballs, and a league with half fastball, half breaking balls.

What I am trying to say is all those correlations were just a slice of the current MLB setting, and hardly something proven universal. What variety of pitches are there, what's the overall defense ability in the league, and probably a lot of other factors would show different correlations.

[Luis_Rivera]

Originally Posted by Skipaway

Also, "findings" like pitchers having little contrl on fly balls actually just means how good the defense is at fielding fly balls.

Skip, you're arguing semantics again.

The overall point is that it's essentially random (with a minor correlation probably due to defense AS I STATED IN THE LINE DRIVE AND GROUNDBALL PARTS). However, it's not random for hitters.

Correlation coefficients aren't about how good the defense is at fielding fly balls... it's about how consistently the defense fields the balls at the same level.

Now if we consider what makes a ball linedrive or flyball, we'd see it's mostly about the spin at contact and contact angle. That's why the kind of pitch a pitcher is throwing might be key. I doubt the result would be the same in a league with mostly fastballs, and a league with half fastball, half breaking balls.

I'm sure theres enough pitchers and pitches within the study to avoid a selection bias like this. Sure, if you want to emulate the earlier parts of the century, then you'd have to do more studies (with different data, however as BIP data would obviously not be available for the early part of the century).

However, that's not the most important thing to me right now... I'm more interested in getting the current MLB environment. Besides, I'm sure the current MLB environment as is described in this study with regards to this is a hell of a lot closer to the MLB environment over the course of history than the current system OOTP has implemented. Therefore, it'd be an improvement.

[Skipaway]

Originally Posted by Luis_Rivera

Correlation coefficients aren't about how good the defense is at fielding fly balls... it's about how consistently the defense fields the balls at the same level.

And I am saying the consistency might only be caused by all study done in a very similar environment. The correlation coefficients showed how consistent the defense is at fielding fly balls, because the defenders are good at fielding the fly balls in the environment it is studied.