THE BOOK--Playing The Percentages In Baseball

Tango - Hang time and initial fielder position are not of much use unless one has accurate hit ball landing locations.  Its pretty clear that the landing locations that we now have are not accurate enough to be useful for the analysis of individual hit balls.  They may be accurate enough to be used in the aggregate for fielding analysis, although Colin is arguing that there is enough bias to make them suspect for that purpose as well.

Why not shoot for the moon a little more and ask for each player to have a (safe) miniature RF transmitter on their uniform, each with a different frequency.  Also build a baseball that also has a few sensors in it that sense things like x,y,z position and velocity, spin, force on contact and have these and the player data transmitted to a data acquisition system which could simultaneously, archive, display and crunch statistical algorithms in real-time.  Only difficult parts would be to get the precision needed on the baseball sensors and possibly building the baseball to have the same characteristics as the non sensored baseball.  But it is all very doable if money is no problem.  Scientists do this with stuff that is much more complex than a baseball game.
vr, Xei

Many years ago, I said we should have GPS on the sweaters / jerseys on all players.

Their wives would be happy…

Xei/2, tracking the movement of the players and the flight of the ball is already being done, by FIELDf/x and TrackMan, respectively.  It’s not a technological problem, and only a small (relatively speaking) implementation challenge at this point.  The problem for us a sabermetric community is that we will not likely have much access to this data.

I’ve never understood the interest in fielder positioning.  We don’t need it all to measure fielding performance—whether a fielder got to a ball (or didn’t) because of where he started or what he did after that doesn’t change the outcome.  Now, if you’re trying to coach players on their fielding, I suppose it would be very helpful to know if your players are using optimal positioning.  But it seems like a secondary concern.

We’ve talked about it in the past.

Does it matter if Pujols’ value comes from HR or BB or doubles?  All that matters is his wOBA.

However, if you can isolate his power skill, his strikezone skill, etc, then we can figure out how to forecast him in the future better.

Same here.  There’s a positioning skill, a “reading the play” skill, a “jump off the bat” skill, a “route” skill, an arm skill, and so on.  What we care about is overall.  But, we need to know each component.

Furthermore, there is more random variation in some components (BABIP) than others, so we want to break it down to the component levels that most closely links to the inherent true skillset of that player.

That’s why we want positioning.

Here’s the Dudek article:

http://www.insidethebook.com/ee/index.php/site/comments/the_first_hardball_times_annual_available_for_download/

Btw, Dudek’s out rate vs hang time curve is roughly consistent with what I’ve seen privately in other data sets.  His analysis was revolutionary at the time, and it stands the test of new information, too.