Comments on: Grandpa Gets All the Calls http://blog.stealingfirst.com/2008/04/10/grandpa-gets-all-the-calls/ Stats, stories and stuff Fri, 29 Aug 2008 07:25:51 +0000 http://wordpress.org/?v=2.6.1 By: Dan Turkenkopf http://blog.stealingfirst.com/2008/04/10/grandpa-gets-all-the-calls/#comment-51 Dan Turkenkopf Wed, 16 Apr 2008 11:38:22 +0000 http://blog.stealingfirst.com/?p=40#comment-51 You think you've got it bad, I'm just an IT consultant. So I've got the programming piece of this down, but I definitely struggle with the analysis. If only I remembered my stats classes better from college. You think you’ve got it bad, I’m just an IT consultant.

So I’ve got the programming piece of this down, but I definitely struggle with the analysis.

If only I remembered my stats classes better from college.

]]> By: Excalabur http://blog.stealingfirst.com/2008/04/10/grandpa-gets-all-the-calls/#comment-50 Excalabur Wed, 16 Apr 2008 08:05:49 +0000 http://blog.stealingfirst.com/?p=40#comment-50 Enh, needs. This stuff is interesting, and I appreciate that people with time to do it do so. I'm a physicist, so I can comment intelligently, but don't have time to do my own baseball research. In ten years or so you could do all kinds of fun stuff with a multivariate regression on age, year, and so on. However, we of course don't want to wait that long. Have fun fixing the selection bias issue, though I believe there are well-known ways to do this (I'm a physicist, so I don't know 'em, we just measure stuff). Enh, needs. This stuff is interesting, and I appreciate that people with time to do it do so. I’m a physicist, so I can comment intelligently, but don’t have time to do my own baseball research.

In ten years or so you could do all kinds of fun stuff with a multivariate regression on age, year, and so on. However, we of course don’t want to wait that long.

Have fun fixing the selection bias issue, though I believe there are well-known ways to do this (I’m a physicist, so I don’t know ‘em, we just measure stuff).

]]> By: Dan Turkenkopf http://blog.stealingfirst.com/2008/04/10/grandpa-gets-all-the-calls/#comment-48 Dan Turkenkopf Tue, 15 Apr 2008 22:16:33 +0000 http://blog.stealingfirst.com/?p=40#comment-48 Thanks, I'll try and take a look at it by year and see what a regression shows. I'm planning on including some graphs for the full version of the article, so hopefully that will meet your needs. Thanks, I’ll try and take a look at it by year and see what a regression shows.

I’m planning on including some graphs for the full version of the article, so hopefully that will meet your needs.

]]> By: Excalabur http://blog.stealingfirst.com/2008/04/10/grandpa-gets-all-the-calls/#comment-47 Excalabur Tue, 15 Apr 2008 20:17:18 +0000 http://blog.stealingfirst.com/?p=40#comment-47 I'd think you'd be better off doing a straight-up binning into one-year bins and then doing a weighted linear regression: there's enough pitchers in the bigs this shouldn't get you too killed. You'd have to weight for number of pitches to avoid the really old guys having too much influence: probably assuming poissionian error would be your best bet. I'm also a big fan of graphs for things like this to see if there's any obvious shape. And yes, of course your selection bias is a big problem: in 10 years you can just look at the same pitchers over time, but that's not an option yet. I’d think you’d be better off doing a straight-up binning into one-year bins and then doing a weighted linear regression: there’s enough pitchers in the bigs this shouldn’t get you too killed.

You’d have to weight for number of pitches to avoid the really old guys having too much influence: probably assuming poissionian error would be your best bet.

I’m also a big fan of graphs for things like this to see if there’s any obvious shape.

And yes, of course your selection bias is a big problem: in 10 years you can just look at the same pitchers over time, but that’s not an option yet.

]]>