Daric Barton? Really? I guess my eyes have deceived me.

Troy Glaus?! That ent's got moves!

Defensive Metrics still remain in the "WTF!?" stage of development, it seems.

Bobby Abreu, NYY  +3

Maybe when baseball was played without any fences...

+/- apparently doesn't dock a first baseman for dropping multiple pop flies.

Utley is such a stud.

Fielding metrics don't account for a lot of what makes a 1B good defensively, but Barton certainly doesn't seem great. Is plus/minus using the same data as RZR and OOZ at Hardball Times?

Defensive metrics are subject to small sample size issues just like offensive metrics are. Anyone also think Paul Bako's .299 EQA at this point invalidates offensive stats.

Glaus rated quite high when he was on the Jays as well.

Watching him play I was actually impressed, he wasn't as bad as I was expecting him to be. That being said, I don't see him being the best in baseball...but I think Glaus is a very under-rated defender.

EDIT: Clearly not under-rated by Dewan of course

Hey Repoz, how about fixing the intro so the page isn't a screwy width?

Defensive metrics are subject to small sample size issues just like offensive metrics are. Anyone also think Paul Bako's .299 EQA at this point invalidates offensive stats.

I'm not disbelieving the metric...it just appears to be in contradiction to my visual observations. But that's okay, my eyes are the lyin' type. I'm interested to see what a larger sample brings.

BTW, does anybody have a quickndirty rule of thumb on how much to regress defensive data?

BTW, does anybody have a quickndirty rule of thumb on how much to regress defensive data?

About twice as much as something like EQA if I recall.

Fielding metrics don't account for a lot of what makes a 1B good defensively

What do you think these other things are?

What do you think these other things are?

Receiving throws, mostly, though I could be mistaken about whether metrics take that into account.

Not sure which metrics cover these and which don't, but this is what I remember of first base defense:

#1 (by far): Saving errors on bad throws. Digging throws out of the dirt, making the right decision on when to come off the base, making tags when possible.

#2-n, probably not in order:
Fielding grounders
Going back on popups
Charging bunts
Starting the 3-6-3/1 double play
Applying tags on pickoffs
Making the unassisted putout to reduce pitcher errors/injuries
Making cutoff plays (positioning and throwing)
Recovering to a reasonable fielding position after holding a runner
Being in position for the pickoff from the catcher
Executing rundowns

#1 (by far): Saving errors on bad throws. Digging throws out of the dirt, making the right decision on when to come off the base, making tags when possible.


I looked at this about 10 years ago and couldn't find any significant 1B effect on the number of errors other infielders had.

This got me to sign up for BillJamesonline.net.

#1 (by far): Saving errors on bad throws.

I followed this a few years ago. there just aren't enough "scoops" (relative to GBs fielded) to make this a huge factor. It's like turning DPs or catching line drives for fielders. Yes, ther eis a skill, but it's margins are small relative to runs saved (say a spread of +2 to -2 for an entire population).

Defensive Metrics still remain in the "WTF!?" stage of development, it seems.

Exactly my reaction.

#1 (by far): Saving errors on bad throws.

I remember years ago reading a blog entry that compared Mo Vaughn and John Olerud in this respect, asserting that there was a huge difference between the two. Olerud's teams always had more errors immediately before and after his tenure with a team, whereas Vaughn, of course, was the opposite.

Defensive metrics are subject to small sample size issues just like offensive metrics are.

try telling that to Darren, Szym.

if you ask me, defensive stats are even more fraught with random variation that offensive ones are.

Utley may be hailed as a good defender, but there's no way he's near the top this year to date. He's got a decent number of errors (and yes I realize that's not a good metric) but they've been of the bone-head variety.

Abreu is a completely different story. #4 sums that up perfectly.

I don't understand how Carlos Gomez is scoring so high on these defensive metrics. I've never seen someone take so many bad routes on balls over his head. He is always taking the wrong step, and then contorting his body to try to get in position. I mean...they're never on TV, but, remember that double JD Drew hit over his head in the 8th on Sunday (on ESPN)? That happens every time someone hits one over his head!!!

i like making fun of carlos gomez, hopefully none of my friends will see this...

Fielding is just as subject to hot and cold streaks as hitting. Glaus is clearly in a "fielding hot streak". He will cool off.
That doesn't make the metric any less valid. It's simply measuring what he's done so far, not making a statement about true talent level. I think thats the problem when a lot of people look at these defensive metrics. Saying Troy Glaus is leading all 3rd baseman in in +/- is not the same thing as saying Troy Glaus is the best 3rd baseman in baseball.

Syzm is saying the exact same thing I am. Paul Bako is hitting for a .299 EQA at this point in time. Thats a fact. Paul Bako's talent level is not that of a .299 EQA hitter. This is probably also a fact.

Carlos Quentin has a higher OPS+ than Manny Ramirez. It's a fact. Carlos Quentin is not a better hitter than Manny Ramirez.

BTW, Glaus currently leads all MLB 3b in RZR and is 3rd in OOZ, and he's tied with Beltre for the best ZR On ESPN (Stats Inc)
So clearly +/- is not out of line here when it comes to Glaus.

i like making fun of carlos gomez, hopefully none of my friends will see this...

The other metrics, RZR, zone rating, also have him at the top. It's not really surprising. He's very fast. That can cover flaws in route running.

I saw Gomez when he came up last year. The kid can cover acres of ground.

I remember years ago reading a blog entry that compared Mo Vaughn and John Olerud in this respect, asserting that there was a huge difference between the two. Olerud's teams always had more errors immediately before and after his tenure with a team, whereas Vaughn, of course, was the opposite.

Sportsjones.

Glaus rated quite high when he was on the Jays as well.

In RZR and stats based off BIS data. Stats Inc's PBP data showed him to still be mediocre.

From what little I've seen of Glaus this year, I'm thinking Glaus's mom scores for BIS. He's been okay, but not great.

#1 (by far): Saving errors on bad throws. Digging throws out of the dirt, making the right decision on when to come off the base, making tags when possible.


I looked at this about 10 years ago and couldn't find any significant 1B effect on the number of errors other infielders had.

I followed this a few years ago. there just aren't enough "scoops" (relative to GBs fielded) to make this a huge factor. It's like turning DPs or catching line drives for fielders. Yes, ther eis a skill, but it's margins are small relative to runs saved (say a spread of +2 to -2 for an entire population).

So ####### what. If there's anything that I learned over the years, it's that absence of evidence DOES NOT indicate evidence of absence.

They're not saying there's evidence of an absence of that skill. They're saying that there are few enough opportunities for that very real skill to be exercised that it isn't particularly relevant.

So ####### what. If there's anything that I learned over the years, it's that absence of evidence DOES NOT indicate evidence of absence.

THat's some terrific reading skill.

Fielding is just as subject to hot and cold streaks as hitting. Glaus is clearly in a "fielding hot streak". He will cool off.
That doesn't make the metric any less valid. It's simply measuring what he's done so far, not making a statement about true talent level. I think thats the problem when a lot of people look at these defensive metrics. Saying Troy Glaus is leading all 3rd baseman in in +/- is not the same thing as saying Troy Glaus is the best 3rd baseman in baseball.

Syzm is saying the exact same thing I am. Paul Bako is hitting for a .299 EQA at this point in time. Thats a fact. Paul Bako's talent level is not that of a .299 EQA hitter. This is probably also a fact.

Sure. But the difference is that it is a FACT that Bako has hit 4 HR this year. Putting aside any possible cases where there was a decision between a hit and an error, it is a FACT that Bako has 30 hits this year. Furthermore, the league-average to which Bako is being compared is a FACT (although the park factors aren't exactly). That says nothing about the predictive utility of Bako's performance this season -- it says everything about how certain we are that Bako has performed to this level this season.

The raw PBP data has the potential to have a LOT of measurement error in it, a good bit of subjectivity. Has Glaus really made X plays that an average 3B wouldn't make (on average)? Maybe a careful review of lots of video data would confirm that but, as it stands, it's far from a FACT. It is a fact that, using Dewan's measurement system (and others based on the same raw data), Glaus comes out well. But our certainty that Glaus really has performed to this level so far this season is not nearly as certain as we are about Bako's performance to date.

That -2 to + 2 range seems low on its face.

If each play made and not made adds up to a little less than a run, we're talking a difference of about 4-5 plays from best to worst in the league?

Looking at the Nats, they've had Nick Johnson all this year, after a full season of Dmitri Young last year. I can assure you that there have been at least twice as many plays as that that NJ has made that Dmitri would've had no chance on.

You can see that in the throwing errors in the infield -- they're way down. But it would show up in other places too. Fielder's choices, for example. Last season, Dmitri couldn't stretch or pick some DP throws that might've completed a DP. It doesn't go down in the scorebook as anything, though.

I'm not sure what methodology was used, but that might be something that would show up in something like Tango's "with or without you" approach. 'Cause the results on their surface don't quite pass the smell test.

it's that absence of evidence DOES NOT indicate evidence of absence.

This has become a very popular catchphrase, but I don't see how it's true. Seems really stupid to me. Of course absense of evidence is evidence of absense. What else could be evidence of absence?

If you do a lot of studying and find no evidence of something, then that is very good evidence of absense. For example, the Loch Ness has been carefully studied and no evidence of the Loch Ness monster has been found. So is that evidence that the Loch Ness monster doesn't exist? Of course it is!

"Evidence" does not mean "100% infallible truth," it just means, uh, "evidence."

It is a fact that, using Dewan's measurement system (and others based on the same raw data), Glaus comes out well. But our certainty that Glaus really has performed to this level so far this season is not nearly as certain as we are about Bako's performance to date.

Fair enough. I can accept that rebuttal. The defense metrics are still not as "proven" as the offense metrics. One question though. ZR as published at ESPN is based on Stats Inc raw data, and RZR is based on BIS Raw Data. Isn't this two different data collection systems, or have I misunderstood?

FWIW, everybody's favorite metric, FRAA, has Glaus at +5 as well. (I know...I know....I shouldn't even quote the number)

I'm not sure what methodology was used, but that might be something that would show up in something like Tango's "with or without you" approach. 'Cause the results on their surface don't quite pass the smell test.

This has been done, actually, both by Rally and MGL. They say pretty much the same thing.

Just taking a glance at MGL's study there, isn't he showing about 12 plays difference between best and worst? That's a bit more sizable than the -2/+2 range from earlier.

12 plays = +3 to -3, as opposed to +2 to -2.

Just taking a glance at MGL's study there, isn't he showing about 12 plays difference between best and worst? That's a bit more sizable than the -2/+2 range from earlier.

It's +6/-3, which is much closer to +/- 2 than it is to +/- 12. Convert that to runs instead of errors, and you're even closer still.

I was too lazy to look, but 9 plays is pretty close to +2 to -2.

I see where you get the +6. Not sure where the -3 is coming from. Karros there is basically -6. Bags and Franco about -4.

Anyway, the range of about 10 plays between best and worst makes sense. It's not sizable, but it certainly shows that there's some skill there.

Also, from his post:"I did not include throws where the first baseman made an error or the runner was safe on a hit (late throw, but no error)."

I'm remembering from last year a number of plays where Dmitri was able to knock a throw down, but wasn't agile enough to make a play for an out. In most cases, the scorer called it a hit, since there was no extra advancement. Now, maybe some of those (almost certainly) wouldn't have been outs even if Keith Hernandez were playing, but certainly a few of the H would've been outs. And if I'm reading that right, those aren't being measured by this metric.

Anyway, the range of about 10 plays between best and worst makes sense. It's not sizable, but it certainly shows that there's some skill there.

In post 17, I said:
Yes, there is a skill, but it's margins are small relative to runs saved (say a spread of +2 to -2 for an entire population).

We might be looking at +/-3. In a given season.

Some of my confusion could be the value of a play.

Isn't the typical play not made worth something in the order of ~.8-.9 runs? Some portion of that's for the extra baserunner, and the other portion for the out not made?

It's a little less than that - .7-.8, and a little less for ROE.

This has become a very popular catchphrase, but I don't see how it's true. Seems really stupid to me. Of course absense of evidence is evidence of absense. What else could be evidence of absence?

Even if I haven't laid eyes on Courtney Love's bong, that doesn't keep me from thinking she might well have one.

Even if I haven't laid eyes on Courtney Love's bong, that doesn't keep me from thinking she might well have one.

Yes, but if you've spent 5 years LOOKING FOR IT, and haven't found it, then that is some evidence that she might not.

Utley may be hailed as a good defender, but there's no way he's near the top this year to date. He's got a decent number of errors (and yes I realize that's not a good metric) but they've been of the bone-head variety.

A bone-headed error doesn't hurt a team more than a normal error, and an error in general doesn't hurt a team (significantly) more than any other ball in his zone that Utley doesn't get to.

That last exchange is a lot more entertaining if you pretend that "bong" is a euphemism for something else.

Yes, but if you've spent 5 years LOOKING FOR IT, and haven't found it, then that is some evidence that she might not.

It could be evidence that his analysis is insufficient to find it. He could be looking for it for 10 years on every golf course in America*. It could mean it doesn't exist, or it could mean he's not looking in the right place.


*The OJ Simpson Method To Find The "Real Killers"™

I feel it is incumbent upon me to point out that every time OJ went to a golf course, the real killer was there.

As such, it's not that bad a search method.