THE BOOK--Playing The Percentages In Baseball

But aren’t errors one of the “most” pitcher/runner/batter independent measures of skill?  I mean, if you want to measure *something* to try to cast light on this question, I think errors are it.

Also, infield singles ... on my crappy softball team, any ball hit towards third base is probably a hit with the bases empty.  In the majors, it’s almost always an out.  It makes sense that there’d be a graduation of skill in between.

Errors are dependent on, I think, in rough order:
- fielding skill
- park
- batter traits (power, speed, GB/FB tendency, error skill)
- pitching traits (GB/FB, error skill)
- other runners on base

While you are probably right about the “most”, I think the park might rank pretty high as well.  I don’t know what kind of caretaking they do in rookie ball, high school, or Israel, but that has to play a part in it.

As for the batter traits, Ichiro and Carl Crawford reach base on error more often than Manny and Frank Thomas (though Mike Piazza is ahead of all of them!) if you look at ROE/(AB-H-K).  They are all fairly close. 

And if a pitcher is a FB pitcher, he’s definitely not going to induce many GB errors.

Finally, since errors are subjective recordings, who knows what standards each human scorer is using.

I’ll grant you the “most”, but I won’t say it’s overwhelming.

Well, I agree with you about all those other factors ... but at the league level, I think they’d cancel out.

I’d bet you that if you looked at error rates (per ground ball, if you prefer), and infield single rates per ground ball, you’d find that they’d be higher in AAA than in the majors, higher in AA than AAA, and so on, all the way down.  I don’t know this is true, but I’d bet.

Just looked at the link in the post, which I should have done in the first place.  Unlike Tango, I find the results fairly convincing, especially the one on DER.

Wouldn’t it be best to use scouting data for something like this?  I’d think so until/unless you had a lot of cross over in players. 

One of my concerns with analyzing the IBL players is sample size.  The top guys are only getting 130-150 PA or so per season, and at 7 innings and 45 games, that’s only 315 innings of pitching/defensive data per team per year.

I like this guy’s methodology also.  And I would guess that he is pretty darn close about the level of the league.

Even if the park was a big factor in error rate, the quality of the park probably correlates well with the quality level of the league anyway.  That might not be true in Israel of course - they might have immaculate fields or they might play on cow pastures.

I reply to this followup:
http://biblemetrics.blogspot.com/2007/10/more-on-league-quality-estimation.html
Plus this:
http://biblemetrics.blogspot.com/2007/10/long-term-league-quality-indicators.html

With the following:

The park for fielders and the runs per game also highly influence errors and steals.

Since parks are not randomly distriuted among rookie ball through MLB, you have a very real bias here.

There’s also potential bias for scoring errors.

As for runs per game, the more the runs, the less likely you are going to steal.  Simply put, the break-even point goes up, as the number of runs scored per game goes up.

Finally, the younger you are, the more you steal.  The aging patterns clearly shows a peak on steals right around 23 years old.  If you have a league filled with 19 and 20 yr olds, of course they will steal alot more.  Furthermore, your strength (throwing arm for catchers) probably peaks much later.

What you are witnessing has alot of biases.  Unless you account for each one, what you are witnessing is not necessarily the result of the one you are hoping for.

You may be proved right in the end, but you haven’t done so yet.

Higher classification stadiums generally have more foul territory, and that could have at least some impact on DER being lower in lower classifications.