Assorted Links

0 recs | Comment 6 comments | Share

Comments

The FIP piece was fantastic

I read through Tango’s original piece in which he develops the coefficients, but I don’t think I ever really understood what was going on. This, I understand. I still like tRA* better, at least in concept, because I’d rather regress stats appropriately rather than all the way to average (which is what tRA* is supposed to do). But I love FIP a little more tonight, and I’ve had a long-standing affair with it. :)
-j

I write at:
Beyond the Boxscore
Red Reporter
Basement-Dwellers.com
Twitter: @jinazreds

by JinAZ on Oct 10, 2009 9:27 PM EDT reply actions 0 recs

Agreed about tRA

The next step would be to create a formula that’s not quite so linear.

by vivaelpujols on Oct 11, 2009 1:15 AM EDT up reply actions 0 recs

With tRA (not so much with the regressed version)...

…you have to ask yourself about the reliability of the underlying batted ball data.

I has a blog. | Ask a sabermagician!

by cwyers on Oct 11, 2009 2:00 AM EDT up reply actions 0 recs

Good point

The difference between a line drive and a fly ball is huge in terms of run and out value. That leaves either using the regressed version, which I don’t think is a bad idea, or using a modified tRA that only considers GB tendencies and lumps LDs and FBs into one.

by vivaelpujols on Oct 11, 2009 2:17 AM EDT up reply actions 0 recs

I thought about that as well

But I wonder if “air balls” would glaze over too much data, since there are huge differences and the distributions are not always the same. Would it consider just an average distribution between LD’s and FB’s if we lump them together?

And Colin, for the non-linearity question, couldn’t we use a BaseRuns model similar to what you did with FIP-BaseRuns, with the BABIP of each batted ball type used as the multiplicative factor for hits and the extra base hit estimator? You still have to figure out how to deal with batted ball data, but it won’t be as linear as FIP/tRA and you get to use the power of BaseRuns, as the article mentions.

Marlin Maniac, a Florida Marlins blog
Come attendIntro to Sabermetrics 101!
Check me out at Beyond the Box Score as well.

by SFiercex4 on Oct 11, 2009 12:11 PM EDT up reply actions 0 recs

There are two ways to do it.

One is to come up with a BsR formula based upon the tRA formula. (It’s fairly straightforward to build a BsR formula from an LWTS formula, which is what tRA is.)

I present another method here.

I has a blog. | Ask a sabermagician!

by cwyers on Oct 11, 2009 12:58 PM EDT up reply actions 0 recs