M’s, Defense, and Statistics
Dave · August 24, 2009 at 7:46 am · Filed Under Mariners
Read this. It’s a fantastic article, and sums up how an organization should go about evaluating these things. It’s not just stats, and it’s not just scouting. It’s both, and they work together.
Comments
13 Responses to “M’s, Defense, and Statistics”
Leave a Reply
You must be logged in to post a comment.
Twisted the knife real well there, but amazingly good write-up.
I remember when Jack Z was hired he said he wanted to build the M’s into a “model franchise.”
Based on this article, it looks like other teams may be following his model very soon.
However, I’m confident Jack and his team can remain a step ahead of the pack.
Excellent article. Two thoughts:
– Interesting that players have more ‘blips’ on defense than offense. Intuitively, it seems to me the opposite would be true
– My only quibble is how the improvement in pitching is minimized. I think this short changes the work Adair et al have done
I think the blips are a result of too many variables, rather than a continued change in ability. Hitting is much easier to measure quantifiably, and thus stays more constant. It is great to hear the M’s number 2 guy talking about sample sizes while showing such knowledge of defensive metrics. Who would’ve thought, even last year?
I really dont understand the Adam Jones comment [Adam Jones is the only one with better all around defensive stats than Gutierrez], when Jones has a -8.9 UZR in center field this year (in 114 games), according to fangraphs. For his career so far it looks like Jones is average or barely above average in center.
Terrific piece by a skilled writer. Thanks for sending this our way. I’d never visited fullcountpitch.com, though I’ll have to make it part of my regular reading now.
MGL (creator of UZR) argues that is something of an illusion. Current offensive stats use outcomes (single, double, etc) that hide interesting data. The bloop “parachute” that drops between the middle infielders, the line drive that skips to a right fielder, and Ichiro’s infield hits, are all classified as “singles” and all formulas like OPS and wOBA that treat them alike are therefore not including information that would otherwise be useful in creating predictive offensive stats. The existence of these “buckets” (single, double, etc), while useful since they record vital information in the game (and for evaluating player records), are nevertheless hiding fuzzier data that might be valuable for prediction. Which makes sense: those fuzzy “blips” in defensive data for hits prevented or allowed should result in corresponding hitting data for outs or hits created. But that gets washed away because all “singles” are the same, offensively.
When we have Hit FX data you can expect to start seeing different “kinds” of singles (etc) just as we now have different kinds of swinging strikes thanks to Pitch FX.
The other thing about the “blips” is sheer sample size.
As MGL says in a Newsweek article addressing Tex:
Recall that it takes something like 500 PA for things like OBP, SLG, and (hence) OPS to become reliable, and that’s most of a season. If defensive stats are accumulated at a slower rate, it follows that you need more than one season’s worth of defensive stats for you to say anything credible with them. (In other words: we don’t have enough major league data about Adam Jones).
Moreover, the data is still subject to random variation. This means even a perfect stat, with perfectly collected data, is going to be wrong for some players some of the time. This is a fact, and it’s an escapable one. There may be limitations inherent to UZR, but even a perfect system wouldn’t be able to avoid this (which is something people who bash UZR, or defensive stats in general, often fail to understand).
Thank you for the great article.
These are also usually the same people that accept BA & ERA as a players worth. They seem to think that if there is any variance, then it can’t possibly be accurate. An argument I heard not long ago was “but they’re never wrong”.
My head almost exploded.
There is some (very brief) discussion in the comments section of this article regarding how much, if any, improved pitching has played in the Mariner’s success.
My own sense is that our pitching has improved simply by not giving as many innings to Silva and Batista (among others) this year. Has anyone seen a piece analyzing the improvement, if any, in the Mariner’s pitching this year?
BTW by “this article” I meant Armida’s piece that was linked to.
It could also appear to be better due to our improved outfield defense. See Washburn, Jarrod.