zMLE Numbers for 2011 Released

These are minor league numbers translated into a MLB context by Dan Szymborski, the same guy who creates the ZiPS projections.

Remember, these are the approximate ML equivalents of 2011 performances, not 2012 projections. Most prospects will have improved over the offseason, and can expect to do at least a little bit better in 2012.

7 months ago Vlad 32 comments 0 recs |

Story-email Email Printer Print

Comments

Is there any information available on the zMLE model?
Couldn’t find it in the file.

by CO_Bucs on Oct 18, 2011 1:46 PM EDT reply actions

Shoot Dan an e-mail: Link.

by Vlad on Oct 18, 2011 2:42 PM EDT up reply actions

Thanks

Link didn’t work, but I’ll try to dig a bit more. I’m not sure how the model works. Do you know if it’s the PECOTA-ish model blend of comparables/similarity scores (B. James) + possible career paths (BP’s Vlad)?

by CO_Bucs on Oct 18, 2011 10:29 PM EDT up reply actions

Projections use comparables and career paths and all the complex stuff.

Translations are just modernized MLEs. James didn’t do pitchers and he didn’t have access to 1977-2006 minor league stats, either at the time, and hadn’t considered BABIP (a lot of the big misses of the old system were high BABIP players).

I haven’t actually used the original MLE (from the 1985 Baseball Abstract) since 1999 or so and I started adding the lowercase z a few years ago to stop from confusing an updated MLE method with the original one. I use the lowercase z since the MLE part is the important part and the z is just a modifier (just like you see bWAR and fWAR for the flavors of WAR).

Now as for ZiPS itself, there’s the very brief explanation from this year’s ESPN Magazine’s Baseball Preview or the longer ZiPS Q&A at BTF.

--
Dan Szymborski
Dan's Stuff is on: BTF, ESPN, Twitter

by D.Szymborski on Oct 19, 2011 1:34 PM EDT up reply actions 1 recs

Bah, second link didn’t work

www.baseballthinkfactory.org/szymborski/zipsqa.rtf

--
Dan Szymborski
Dan's Stuff is on: BTF, ESPN, Twitter

by D.Szymborski on Oct 19, 2011 1:35 PM EDT up reply actions

Thanks a lot, Dan

I really appreciate you taking the time to address the question and provide some useful information.

by CO_Bucs on Oct 19, 2011 11:04 PM EDT up reply actions

After reading the Q&A, it is definitely a lot clearer. And, I really like your remarks on BABIP and how that contributed to mistakes in evaluation.

My next question is a hypothetical and long-winded:
It’s the present day and I’m the VP of Baseball Operations for MLB Team X. Since most, if not all, teams develop their own metrics and/or evaluation/projection systems (forgive me, I read this somewhere), what separates ZiPS (and other freely available tools) and BP’s proprietary tools from an organization’s own tools that are being developed internally. (Note: that is possibly an ignorant statement because it implies a greater accuracy than no-cost tools)

Again, my assumption is that a ZiPS/PECOTA/zMLE (I know they are different tools) are not as “robust” as something that an MLB club is developing and implementing in day-to-day operations. That may be faulty assumption.

To put it another way…. if you had unlimited resources (funds, research assistants, and processing capacity), what data would you 1) want to collect (as an input in a modified existing tool), 2) what “new data” is not currently being collected or undercollected, 3) what new metrics might you be interested in, and 3) what other academic disciplines might be drawn on to improve this type of analysis (e.g., spatial statistics).

Apologies for the long-winded response/question.

by CO_Bucs on Oct 19, 2011 11:22 PM EDT up reply actions

Most teams’ focus isn’t so much on projection tools and the like, but database experts. The ease of getting the information to the people who need them is generally the priority, rather than fancy new metrics. The projection field is one where it’s useful to have people with experience with working with the specific kind of data as it’s a pretty big undertaking to build something from scratch. I can’t talk too much about it, because non-disclosure agreements are pretty strict (though I haven’t done anything since I’ve started working for ESPN as there’s a conflict of interest). Sadly, you’ll probably get the same response from the other projection guys as they’re pretty much in the same baot.

I’d like to really accumulate as much defensive and positioning data as possible. I don’t think at this point, there’s not a WHOLE lot that’s really undercollected, I’d want my organization to be focusing on how we apply the information out there and how to evaluate non-statistical eras in a more objective fashion.

--
Dan Szymborski
Dan's Stuff is on: BTF, ESPN, Twitter

by D.Szymborski on Oct 20, 2011 12:29 AM EDT up reply actions

Interesting points, thanks for the feedback. Take home message seems to be focused on implementing the previously built databases and models, rather than a more “academic” approach of building, building, building. I remember reading BP last summer – they introduced and retired a new metric about six weeks apart (if I remember correctly, it was a derivative of SIERA).

At the organization level, it’s easy to understand the point that resources are better spent at implementation/delivery/application. But, I suppose there’s always inefficiencies in what data are being collected (what’s “popular”) and the different lag times throughout MLB clubs at using the data.

Your point on defensive/positioning data was excellent. With respect to your point on evaluating the non-statistical eras, does that mean an effort to build better databases for comparable/similarity models or is it to better understand the successes/failures of the layman’s understanding of “traditional scouting”? Or, is it something else.

by CO_Bucs on Oct 20, 2011 1:49 AM EDT up reply actions

OT question:

Is anyone else having trouble with the SB Nation iPhone app? I will look at a story and then go to the comments and they have nothing to do with the story I was reading.

You gotta aim high to fail so big. - Trace Beaulieu

by IAPiratesFan on Oct 18, 2011 5:06 PM EDT reply actions

minor quibble

Players shouldn’t necessarily be expected to do better next year. Just like major leaguers (even young ones) have better and worse years, so do minor leaguers. Players can easily perform worse even as their skill level increases.

by epoc on Oct 18, 2011 6:49 PM EDT reply actions

They won’t always improve, but a player young enough to be considered a “prospect” will improve more often than not. Just not always enough to keep up with the competition…

by Vlad on Oct 18, 2011 7:07 PM EDT up reply actions 1 recs

right

Prospects will improve. What I’m saying is that their numbers (even MLEs) won’t necessarily improve, because of random variation. So while you can expect prospects to be better next year, you can’t really expect them to “do” better.

by epoc on Oct 18, 2011 7:36 PM EDT up reply actions

Hot damn!

I’ve been waiting for the zMLE numbers.

パトリック

by patthatt on Oct 18, 2011 7:34 PM EDT reply actions

It's bad enough that we've got 1,234,345,756,354,675 more baseball stat acronyms to remember...

but now there’s these obnoxious combinations of upper and lowercase letters in them!!

by impliedi on Oct 18, 2011 10:35 PM EDT up reply actions

The lowercase letter make them look more fancy

by CO_Bucs on Oct 19, 2011 1:06 AM EDT up reply actions

It's a joking reference to DIPS

He published his first projections on BPro about the same time that Voros was rolling out his research on the same site.

Stands for (s)Zymborski Projection System, if I recall correctly.

by Vlad on Oct 19, 2011 8:22 AM EDT up reply actions

But then...

shouldn’t both the “z” and the “i” be lowercase?? It really doesn’t make a lot of sense, if ya ask me.

by impliedi on Oct 19, 2011 11:34 AM EDT up reply actions

The “Z” is capitalized because it’s the start of a proper noun.

by Vlad on Oct 19, 2011 11:55 AM EDT up reply actions

I was just kidding the with fancy, we should be thankful for all the data

by CO_Bucs on Oct 19, 2011 11:06 PM EDT up reply actions

What about THESE ZiPS projections???

by impliedi on Oct 18, 2011 10:37 PM EDT reply actions

Over Macho Grande? No, I'll never get over Macho Grande.

by white angus on Oct 19, 2011 8:17 AM EDT up reply actions

still waiting for the GEORGE ZIP reference

dammit!

by white angus on Oct 19, 2011 5:41 PM EDT up reply actions

just for fun

here are the pirates minor league players with a higher OPS than lyle overbay according to zMLE (min 150 ABs):

Matt Curry
Dusty Brown
Josh Harrison
Jordy Mercer
Alex Presley
Matt Hague
Ramon Cabrera
Jose Osuna
Starling Marte
John Bowker
Robbie Grossman
Alex Dickerson
Adalberto Santos

by johnnycuff on Oct 19, 2011 11:37 AM EDT reply actions

And what's also funny is that less than half of those guys will ever see a starting MLB job, let alone have one for ten years.

by impliedi on Oct 19, 2011 11:46 AM EDT up reply actions

are we still slamming Overbay on here?

whom is going to be the 2012 whipping boy for the Pirates?

a. 2009 whipping boy Cedeno, working hard to get his title back
b. pedro (insert any last name here)
c. a soon to be aquired player who is nicknamed “garbage”, “trash”, “worthless”

by white angus on Oct 19, 2011 2:23 PM EDT up reply actions

whom is going to be the 2012 whipping boy for the Pirates?

If he didn’t want to be criticized for his play, he should have played better.

Nobody is trying to say that he’s a bad person – just that at this point in his career, he’s no longer a starting-caliber ballplayer.

by Vlad on Oct 19, 2011 3:16 PM EDT up reply actions

and we all know this already... he was let go 3 months ago.

we’ll be bitching about Overbay until someone else takes his spot on the whipping post.

by white angus on Oct 19, 2011 5:42 PM EDT up reply actions

if it makes you happier

substitute brandon wood for overbay’s name. or matt diaz. or josh harrison. or any player on any team with a lousy OPS. i don’t really care. i did say this was just for fun after all.

by johnnycuff on Oct 19, 2011 5:50 PM EDT up reply actions