Comments on: Ichiro, Ichiro, Ichiro

By: naviomelo

naviomelo — Wed, 12 Nov 2008 00:24:22 +0000

Thanks Mat. I wasn’t trying to dismiss defensive metrics, but some of the results just seemed like oddities to me based on what our common perceptions are of the players in question.

By: msb

msb — Mon, 10 Nov 2008 16:19:40 +0000

but Derek! I thought we only looked at numbers and nothing else, blindly ignoring real baseball.

I’m so confused.

By: DMZ

DMZ — Mon, 10 Nov 2008 14:52:17 +0000

They’ve diverged greatly before on players and positions. Like all stats, it’s important to look at them with open eyes and knowing their limitations. With defensive stats, which are still the roughest of our statistical tools, even more so.

By: philosofool

philosofool — Mon, 10 Nov 2008 13:51:05 +0000

This is worrisome news for the idea that PMR and UZR are actually measuring defensive performance. If these metrics are able to disagree so wildly about the very same performance, why do we think that they are measuring the same thing, viz. defensive performance?

By: vj

vj — Mon, 10 Nov 2008 10:02:41 +0000

The link doesn’t go directly to the comment but the thread’s start. I am referring to comment #318. {BTW the edit function isn’t working for me. I am on IE 6 here at work.}

By: vj

vj — Mon, 10 Nov 2008 09:58:44 +0000

As far as I know, Erstad has a pretty good reputation as a defender and UZR has liked him. See for instance this comment by tangotiger in last year's free agent thread at insidethebook.

By: Mat

Mat — Mon, 10 Nov 2008 09:30:17 +0000

Erstad in CF for the Astros in 2008.

Erstad played a measly 304 innings in CF for Houston. Erstad had 97 putouts, so a 105.93 would mean that PMR had him at roughly 91-92 expected putouts. It may be somewhat improbable for Erstad to have been a handful of plays above average, but given such limited playing time, it’s hardly a crazy result.

Jamie Burke had an OBP 7% better than league average in 2007–can we take OBP seriously? (Sorry, I know you left open the possibility for small sample size as an explanation, but the general willingness to dismiss defensive stats over a few stray data points is a sore spot for me. The same thing happens with offensive stats all the time and we’re willing to take those seriously–sometimes much too seriously over small sample sizes.)

By: DMZ

DMZ — Mon, 10 Nov 2008 06:53:15 +0000

Which Erstad?

By: naviomelo

naviomelo — Mon, 10 Nov 2008 06:42:51 +0000

According to the comments, they rated Darin Erstad way above average at 105.93 in CF. Can we take PMR seriously, or is this just small sample size theater noise?

By: marc w

marc w — Mon, 10 Nov 2008 02:54:36 +0000

Skyking -

That’s really great news. I’ve worried that a lot of the discrepancy between the defensive metrics is due to the different ways zones/balls are classified. The OF rankings (and 3B ones, to a lesser degree) don’t correlate well – MGL’s work should clear up a lot of questions.