Justify why Josh Hamilton should be in the Top 10 in AL MVP

[Rob]

I love this thread... a classic badnews freakout at its best. But I can't believe nobody caught this at the time...

 

Park factors are among the most overrated "stats" among statheads. Non-statheads don't even think about them.

 

The use of statistical formulas in this regard is to normalize the data so that someone can make a judgment between two or more players that play in different environments. However, the SD for park factors is so small that most of the time the data have very little meaning. Outside of places like old Coors (before the humidor) and Petco/Comerica most places play relatively similar.

 

The question that's important is, does player A perform better in park 1 than he would in park 2? A guy like Slappy was able to take advantage of Coors and Pro-Player because they had huge outfields. He really is/was a better player when he played in Coors b/c the park suited his abilities. A guy who drives the ball in the gaps isn't going to be as good in Pro-Player as he would in a more symmetrical park.

 

A guy like Hamilton who mashes the snot out of the ball will still mash the snot out of the ball anywhere he plays. So that means he's very likely to play well in a hitter friendly environment and play well in a pitcher friendly environment.

 

All and all I would just look at OPS+ to try to determine offensive output when comparing players.

 

OPS+ is OPS adjusted by league and (gasp!) park factors. Please dear god stop pretending to be a stats guy.

[webjim25]

I've never laughed so hard on a message board in my life. My God this was funny. I had a few colorful movie references that I thought "this thread was like," but since those zingers were dropped a couple pages back, I'll keep em in the bag for next time.

[Mephistopheles]

I love this thread... a classic badnews freakout at its best. But I can't believe nobody caught this at the time...

 

Park factors are among the most overrated "stats" among statheads. Non-statheads don't even think about them.

 

The use of statistical formulas in this regard is to normalize the data so that someone can make a judgment between two or more players that play in different environments. However, the SD for park factors is so small that most of the time the data have very little meaning. Outside of places like old Coors (before the humidor) and Petco/Comerica most places play relatively similar.

 

The question that's important is, does player A perform better in park 1 than he would in park 2? A guy like Slappy was able to take advantage of Coors and Pro-Player because they had huge outfields. He really is/was a better player when he played in Coors b/c the park suited his abilities. A guy who drives the ball in the gaps isn't going to be as good in Pro-Player as he would in a more symmetrical park.

 

A guy like Hamilton who mashes the snot out of the ball will still mash the snot out of the ball anywhere he plays. So that means he's very likely to play well in a hitter friendly environment and play well in a pitcher friendly environment.

 

All and all I would just look at OPS+ to try to determine offensive output when comparing players.

 

OPS+ is OPS adjusted by league and (gasp!) park factors. Please dear god stop pretending to be a stats guy.

 

Wow, I laughed so hard I teared up a bit. Thanks for bumping this thread. I can't believe I haven't seen this.

 

Being the resident stat guy, I think this is a good opportunity for me rant on park factors.

 

Everyone seems to misinterpret them. It doesn't matter how statistically inclined you are. It's not because people are dumb (they're not). It's because they misinterpret the meaning of them. OPS+ is NOT a measure of talent. It never was intended to be. EqA is NOT a measure of talent. It was never meant to be. GPA is NOT a measure of talent. It was never meant to be. DRA is NOT a measure of talent. It was never meant to be. wOBA is NOT a measure of talent. It was never meant to be.

 

All of these major statistics measure production. The difference between the analysis of talent and production is the difference between the future and past. A players output is not a martingale, if it were the two would be the same. Production has an intrinsic concept of being compared to something more than talent would. Statistics measuring production always boil down being compared to one of two things: average level players and replacement level players.

 

Since they want to measure a players production rate to the rate of an average (replacement) level player. The analysis is relatively simple. What would a league average (replacement) level player due if he was given the playing time of our player? Just the league average run effect. HR, BB, 2B, 3B, K, etc PFs are irrelevant. We don't care about how the park plays on a micro level. The only thing that matters is what happens to an average player on a macro rate.

 

However, when trying to make some educated guess on a players true talent level, a lot more work is needed to be done. This is why things like EqA, wOBA, etc suck ass at this. Parks effects on an individuals stat line are much more of a micro level than a macro effect. To make changes like this we need a large spectrum of park factors which eventually is more of a dynamical system than any kind of park factor we're used to seeing. In order to make guess worth crap on a true translation out of a park for a player we need to a stat line completely. We can't just make a simple change.

 

I don't have the answer to this. That's the million dollar question. No one has a system that does this well. The best may be the DTs on BP, but they're still useless. Quite frankly, none of the people interested in these sorts of things have the mathematical modeling and probability background in order to derive something better. Let's be honest, most people working in sabermetrics really don't have more than an undergraduate college background in stat or whatever. It's pretty damn obvious reading their "research". That's not to say they're dumb or wrong or anything. It's just that if someone with relevant background got interested, things would get real interesting.

 

But anyways, about PFs and OPS+. It measures production. It doesn't measure talent. Parks affect each player differently, we need systems that adjust for that.

 

 

/rant.

[Rob]

I actually recall hearing that John Dewan was working on adding a sort of hitters database into his fielding work... tracking velocities, vectors, etc... the next logical step would obviously just be slapping together a computer model with stadium dimensions and weather patterns... it could lead to some very interesting studies.

[Mephistopheles]

The problem is that the studies will still be geared towards production analysis, not talent analysis. They'll try to make a estimation of park neutral production. It's a step in the right the direction. I'm relatively skeptical of exact vector analysis myself. In order to get the best measure of true talent in a park neutral setting you're going to need to move to a probability based assessment.

 

Theoretically, you can start out by analyzing the vectors forget the actual outcome of the event. Then assign a probability to every possible event based on the vector. This is a good start. However, what you'll find that is that these vectors probably have little predicative worth. So we'll get the point to having probabilities of distributions.

[Rob]

Believe it or not, I'd actually had all this occur to me before. I'm just not smart enough to figure out the next step.

[Guest]

I love this thread... a classic badnews freakout at its best. But I can't believe nobody caught this at the time...

 

Park factors are among the most overrated "stats" among statheads. Non-statheads don't even think about them.

 

The use of statistical formulas in this regard is to normalize the data so that someone can make a judgment between two or more players that play in different environments. However, the SD for park factors is so small that most of the time the data have very little meaning. Outside of places like old Coors (before the humidor) and Petco/Comerica most places play relatively similar.

 

The question that's important is, does player A perform better in park 1 than he would in park 2? A guy like Slappy was able to take advantage of Coors and Pro-Player because they had huge outfields. He really is/was a better player when he played in Coors b/c the park suited his abilities. A guy who drives the ball in the gaps isn't going to be as good in Pro-Player as he would in a more symmetrical park.

 

A guy like Hamilton who mashes the snot out of the ball will still mash the snot out of the ball anywhere he plays. So that means he's very likely to play well in a hitter friendly environment and play well in a pitcher friendly environment.

 

All and all I would just look at OPS+ to try to determine offensive output when comparing players.

 

OPS+ is OPS adjusted by league and (gasp!) park factors. Please dear god stop pretending to be a stats guy.

Are you back on the meds?

 

You've taken the post completely out the context of the other posts.

 

To make it make sense to you I'll add:

 

All and all I would just look at OPS+ to try to determine offensive output when comparing players, even thought park effects are overstated.

Edited April 23, 2009 by CubinNY

[Guest]

 

But anyways, about PFs and OPS+. It measures production. It doesn't measure talent. Parks affect each player differently, we need systems that adjust for that.

 

I'm not sure I follow you all the way. "True talent", if such a thing actually exists (and I have grave doubts) should be expressed in production (including long flyballs outs, gappers, and slap hits) not some mythical internal and hypothetical variable like potential.

 

I agree with what I've bolded above but I think it's is possible to place players in somewhat arbitrary categories based on their production.

 

For example, Pierre made his career playing in big parks like ProPLayer and Coors. Those parks probably suited his "true talent" better than Wrigley. A guy who pulls the ball is going to mash at places that have shorter dimensions down the line, etc.

 

The only way to know that is to look at what actually happened.

[Derwood]

Never did get that badnews Top 10 list.