Predictive / Descriptive

[tangotiger]

In response to some questions here:

Suppose you hit a no-man's land ball: 70 mph, high launch angle, that lands 220 feet from home, halfway between the IF/OF.  That is, essentially, a .900 xBA.

Suppose you hit a hard ball to straight away CF: 100 mph, medium launch angle, that would have landed 340 feet from home, but the CF had plenty of time to get under it.

DESCRIPTIVELY, the first one is far far better.

PREDICTIVELY, the second one is far far better.

The first one is, basically, a bad swing, that the batter lucked into falling between all the fielders.

The second one is, basically, a good swing that the batter bad-lucked into getting just a bit too under the ball.

So, do we want that first play, the texas leaguer, to show a .900 xBA or a .100 xBA?

Do we want that second play, the deep CF hit, to show a .100 xBA or .600 xBA?

I'll guarantee you: we won't get consensus.  So, I specifically made it descriptive, but limited it to the two variables most in the batter's control (speed, angle). It's not fully descriptive, just mostly.

 

Edited July 9 by tangotiger

[Frobby]

23 minutes ago, tangotiger said:

In response to some questions here:

Suppose you hit a no-man's land ball: 70 mph, high launch angle, that lands 220 feet from home, halfway between the IF/OF.  That is, essentially, a .900 xBA.

Suppose you hit a hard ball to straight away CF: 100 mph, medium launch angle, that would have landed 340 feet from home, but the CF had plenty of time to get under it.

DESCRIPTIVELY, the first one is far far better.

PREDICTIVELY, the second one is far far better.

The first one is, basically, a bad swing, that the batter lucked into falling between all the fielders.

The second one is, basically, a good swing that the batter bad-lucked into getting just a bit too under the ball.

So, do we want that first play, the texas leaguer, to show a .900 xBA or a .100 xBA?

Do we want that second play, the deep CF hit, to show a .100 xBA or .600 xBA?

I'll guarantee you: we won't get consensus.  So, I specifically made it descriptive, but limited it to the two variables most in the batter's control (speed, angle). It's not fully descriptive, just mostly.

 

Thanks very much for coming on and posting this explanation.   I have two follow-up questions:

1.   Do you think (or have you studied whether) the difference between wOBA and xwOBA (or BA/xBA) is mostly the result of luck, or are there other reasons that mostly or partially explain why there is a differential?  For example, do pull hitters tend to exceed their xwOBA because the fences are shorter in the areas they tend to hit the ball, compared to hitters who tend to hit towards CF?

2.  Do you think (or have you studied whether) batters have more control over launch angle than they do over the direction the ball travels from foul line to foul line?

Thanks in advance for any answers you are able to give.

Edited July 9 by Frobby

[tangotiger]

The reality is that the truth is somewhere in the middle of wOBA and xwOBA.  And you can find some of the reasons based on whether a batter is getting alot of distance or not, or hitting the ball hard or not.

And no, spray tendencies do not help to close the gap for the most part.  In these cases, the few exceptions prove the rule.

[btdart20]

Is this Tom Tango?  If he talks like a tangotiger, he must be tangotiger!

How long is the ball in the air on either hit?  Squaring up a ball to "dump" it over the IF might not be a Texas leaguer, but the less time the ball is in the air is a predictor.  If speed and angle are variables of the quality of contact then, the quality of the 2 strike/Cobbsian approach matters.

[tangotiger]

34 minutes ago, btdart20 said:

How long is the ball in the air on either hit?  Squaring up a ball to "dump" it over the IF might not be a Texas leaguer, but the less time the ball is in the air is a predictor.  If speed and angle are variables of the quality of contact then, the quality of the 2 strike/Cobbsian approach matters.

The time in the air is DESCRIPTIVE, but NOT PREDICTIVE.

The Arraez/Ichiro hitter is an exception.  It's how you know the rule works, because the exceptions are so obvious and limited, so myopic.

The power of a batter is highly predictive.  The launch angle is highly predictive.

It's not just me that's found this.

[Sports Guy]

4 minutes ago, tangotiger said:

The time in the air is DESCRIPTIVE, but NOT PREDICTIVE.

The Arraez/Ichiro hitter is an exception.  It's how you know the rule works, because the exceptions are so obvious and limited, so myopic.

The power of a batter is highly predictive.  The launch angle is highly predictive.

It's not just me that's found this.

So how would you look at a guy like Colton Cowser?

He has an above average launch angle, high exit velos, lots of barrels, good chase rate…but then he whiffs a lot and gets a lot of Ks.

His expected stats are much better than his actual stats.

How much bad luck is involved vs the hole in his swing, especially on off speed pitches?

[Warehouse]

Extremely informative! Thank you. 

For me, the discussion of whether xwOBA is more descriptive vs. more predictive is somewhat beside the point. What I care about is whether wOBA or xwOBA is more predictive.  From what I can tell based on various studies, the answer seems to be xwOBA is more predictive for sample sizes of less than a season and wOBA is more predictive for sample sizes of a season or more.

[Perhaps the answer is that ZIPS or Steamer projected wOBA is the most predictive metric, but using black box model outputs is often less than satisfying in exchanges with other fans].

[btdart20]

1 hour ago, tangotiger said:

The time in the air is DESCRIPTIVE, but NOT PREDICTIVE.

The Arraez/Ichiro hitter is an exception.  It's how you know the rule works, because the exceptions are so obvious and limited, so myopic.

The power of a batter is highly predictive.  The launch angle is highly predictive.

It's not just me that's found this.

That makes sense.  Air time is an output of the inputs.  I agree they are highly predictive.  Measure the right inputs to chase your goals.

I guess where I'm going is why would Arraez or Ichiro be such outliers?  I know the game has shifted a lot over the past 40-50 years.  But the predictive average talent level of players is so much higher than it's ever been while descriptive  BA and OBP seem to be going lower and lower pursuing SLG.  I know I'm introducing the pitching variables into the discussion, so it's a bit off the general point.

We've got a cloning machine for power/speed athletes with plyometrics and quick twitch training.  But not Tony Gwynn.  Is it as simple as nature vs. nurture for the batting eye and hit tool?  What's the key to that leap?

[Since1984]

4 hours ago, tangotiger said:

In response to some questions here:

Suppose you hit a no-man's land ball: 70 mph, high launch angle, that lands 220 feet from home, halfway between the IF/OF.  That is, essentially, a .900 xBA.

Suppose you hit a hard ball to straight away CF: 100 mph, medium launch angle, that would have landed 340 feet from home, but the CF had plenty of time to get under it.

DESCRIPTIVELY, the first one is far far better.

PREDICTIVELY, the second one is far far better.

The first one is, basically, a bad swing, that the batter lucked into falling between all the fielders.

The second one is, basically, a good swing that the batter bad-lucked into getting just a bit too under the ball.

So, do we want that first play, the texas leaguer, to show a .900 xBA or a .100 xBA?

Do we want that second play, the deep CF hit, to show a .100 xBA or .600 xBA?

I'll guarantee you: we won't get consensus.  So, I specifically made it descriptive, but limited it to the two variables most in the batter's control (speed, angle). It's not fully descriptive, just mostly.

 

Reminds me of how many flares Jeter hit in his career