Castro and Theo to meet Friday

[sneakypower]

aside from failing to account for the situational run probability as well as the skill level of the baserunner, it also fails to account for the most important piece of the puzzle: the fielder! particularly, where they are standing at the time the ball is hit. this can be as intricate as using the pitch f/x system to also pay attention to the activity of the fielders (which, as far as i know, it doesn't, and is probably the biggest flaw to the system), but even accounting for the defensive field the player is in would be a great start. preventing a runner from advancing to third is easier, all things being equal, when you're on the left side of the field.

 

after thinking about this a little bit i actually think that uzr does account for position, at least it should, since it separates of'ers by position. i'm not sure if they're isolating position for context or relying solely on park factors here, but whatever it is IT'S NOT WORKING GUYS.

yes, UZR does account for position - so Matt Holliday fares well not necessarily even because he's a good fielder, but rather he's good compared to Ibanez, Bay, Willingham, etc.

 

but then that's all accounted for in positional adjustment

[UK1679666180]

I've been saying UZR is no good for at least three years, so I'm ahead of the curve on that one. It's unreliable year to year and does not account for all variables. It's not valid. People should stop using it as an argument anchor (but they won't).

 

*pats self on back

 

5 and a half years here.

 

 

Then what is a reliable way to evaluate defense? Scouting?

 

Scouting via play by play data.

 

ok so my question is does one need a hawaiian shirt and binoculars to scout the play by play data?

 

jk i just hate baseball but no seriously man, do you really believe that the existing information captured by play-by-play data tells us as much as we need to know about a defensive player's abilities and execution on the field?

 

Never seen binoculars with a scout and very seldom a hawaiian shirt.

 

I think you can, its the most accurate way to do so. The technology is there to analyze the ball coming off the bat, they can on a play by play basis analyze the read, break, and speed of the defender as far as long it took him to get from point a to point b. You can gauge the velo of the throw either by radar gun or naked eye as well as the accuracy and how long it took him to get rid of it.

 

Play by play data incorporating defense is what Oak has tried with their recent market ineff. approach.

[UK1679666180]

Fwiw, I've done freelance work for Stats Inc doing video analysis (baseball only), I've seen their technology and how its applied, for defense as well as other aspects, it should be the present as far as data analysis.

[seanimal]

aside from failing to account for the situational run probability as well as the skill level of the baserunner, it also fails to account for the most important piece of the puzzle: the fielder! particularly, where they are standing at the time the ball is hit. this can be as intricate as using the pitch f/x system to also pay attention to the activity of the fielders (which, as far as i know, it doesn't, and is probably the biggest flaw to the system), but even accounting for the defensive field the player is in would be a great start. preventing a runner from advancing to third is easier, all things being equal, when you're on the left side of the field.

 

after thinking about this a little bit i actually think that uzr does account for position, at least it should, since it separates of'ers by position. i'm not sure if they're isolating position for context or relying solely on park factors here, but whatever it is IT'S NOT WORKING GUYS.

yes, UZR does account for position - so Matt Holliday fares well not necessarily even because he's a good fielder, but rather he's good compared to Ibanez, Bay, Willingham, etc.

 

but then that's all accounted for in positional adjustment

 

i was speaking to the arm component of uzr, not the whole of uzr. i should have made that clear but i just wrote uzr cause i was thinking about at work and needed to spit it out.

[seanimal]

Never seen binoculars with a scout and very seldom a hawaiian shirt.

 

well but what about this guy

 

http://www.chicagotribune.com/media/alternatethumbnails/story/2011-09/23456406441600-20145204.jpg

 

I think you can, its the most accurate way to do so. The technology is there to analyze the ball coming off the bat, they can on a play by play basis analyze the read, break, and speed of the defender as far as long it took him to get from point a to point b. You can gauge the velo of the throw either by radar gun or naked eye as well as the accuracy and how long it took him to get rid of it.

 

Play by play data incorporating defense is what Oak has tried with their recent market ineff. approach.

 

well sure, all baseball data is accumulated through observation. my issue is that, to my knowledge, there is no standardized system in place to accurately record EVERY activity on the baseball field a la pitch f/x (but not just for pitches). while we can certainly watch video and observe a player's reaction to a batted ball and how long it takes that fielder to throw the ball from the moment he receives it, so on and so forth, that activity is not quantified in a standardized way to allow for comparison and analysis amongst an entire field of other comparable data points. furthermore, gauging anything by radar gun, and especially the naked eye, simply doesn't provide the level of accuracy and comparability required. those tools might serve the purpose of a scout, but for those looking for information that isn't colored by the fallibility of human observation, it simply is not good enough.

[UK1679666180]

Never seen binoculars with a scout and very seldom a hawaiian shirt.

 

well but what about this guy

 

http://www.chicagotribune.com/media/alternatethumbnails/story/2011-09/23456406441600-20145204.jpg

 

I think you can, its the most accurate way to do so. The technology is there to analyze the ball coming off the bat, they can on a play by play basis analyze the read, break, and speed of the defender as far as long it took him to get from point a to point b. You can gauge the velo of the throw either by radar gun or naked eye as well as the accuracy and how long it took him to get rid of it.

 

Play by play data incorporating defense is what Oak has tried with their recent market ineff. approach.

 

well sure, all baseball data is accumulated through observation. my issue is that, to my knowledge, there is no standardized system in place to accurately record EVERY activity on the baseball field a la pitch f/x (but not just for pitches). while we can certainly watch video and observe a player's reaction to a batted ball and how long it takes that fielder to throw the ball from the moment he receives it, so on and so forth, that activity is not quantified in a standardized way to allow for comparison and analysis amongst an entire field of other comparable data points. furthermore, gauging anything by radar gun, and especially the naked eye, simply doesn't provide the level of accuracy and comparability required. those tools might serve the purpose of a scout, but for those looking for information that isn't colored by the fallibility of human observation, it simply is not good enough.

 

You'll never get that point though, even with pitchers and hitters. There's too much luck and variables even with the more advanced data with pitching and hitting. The goal is to get the best with what you have and right now, defense still requires that human element.

[Guest]

Sounds like this meeting just happened since Sveum just said he just met Castro on WSCR.

[seanimal]

You'll never get that point though, even with pitchers and hitters. There's too much luck and variables even with the more advanced data with pitching and hitting. The goal is to get the best with what you have and right now, defense still requires that human element.

 

I disagree. If we applied a modified pitch f/x system that used high-speed video to track the movement of every moving part on the field, we absolutely could get to the point. We would know the precise speed and trajectory of any ball, we would know the fielder's position at the time of the hit, we would know how long it took them to not only break to the intercept but also how long it took for them to receive and transition to the throw, so on and so forth. for every single play. no more relying on human subjectivity, no more guessing, no more unknowns.

[woodchip2153]

You'll never get that point though, even with pitchers and hitters. There's too much luck and variables even with the more advanced data with pitching and hitting. The goal is to get the best with what you have and right now, defense still requires that human element.

 

I disagree. If we applied a modified pitch f/x system that used high-speed video to track the movement of every moving part on the field, we absolutely could get to the point. We would know the precise speed and trajectory of any ball, we would know the fielder's position at the time of the hit, we would know how long it took them to not only break to the intercept but also how long it took for them to receive and transition to the throw, so on and so forth. for every single play. no more relying on human subjectivity, no more guessing, no more unknowns.

 

They are working on this, called field f/x. I may be getting it confused with hit f/x but I believe they have it at Pac Bell and maybe one or two other stadiums but only mlb has access to the data.

 

Until they make this info available, the best we can do use UZR, +/-, and scouts create a best estimate of a players defensive value.