Minnesota's Mounting Defensive Woes

One of the biggest problems with the team in the last few years is their inability to match their pitchers with their outfield. They haven't seemed to par their flyballers with the best defensve fielders, so mediocre pitching looks downright awful. Ricky Nolasco and Vance Worley weren't great, but they were also BABIPed to death 

SDI is bunk, unfortunately.

SDI adds the results from other defensive systems together -- heck just about every defensive system you can think of is taken and all added together to form SDI. 

Some systems count some things that others do not count, but there is considerable overlap. Any stat that has overlap (or is counted in multiple systems) will thus be counted multiple times by the SDI index. 

For example, with SDI, one error is suddenly counted several times inside of various algorithms that are trying to put forth a theory as to what an error is actually worth. So rather than SDI figuring out what an error is ACTUALLY worth (picking the best theory), SDI simply takes every definition of an error's worth and adds them all together. Suddenly an error's value is inflated into the stratosphere.

So if a player happens to be outside the norm on some trait that is counted in multiple systems, SDI explodes that value out. But what if a great, important trait is only counted in one or two systems, and the player excels at it? It's buried in the mess.

SDI might be the most comical baseball statistic ever and I'm a bit amazed that the SABR guys aren't embarrassed by its existence. 

One of the things about that Def rating (which is Fangraphs' Defensive Runs Above Average) that really sticks out to me is how much it hated Max Kepler's D last year. Nice to see SDI views him in a more positive light.

 

I agree that we should be able to expect a better performance from him in '17. He played more innings in RF last season with the Twins than he had his entire minor league career. 

The DEF rating at Fangraphs takes into account position.  It has a positional adjustment. It's why only 3 qualifying RF have a positive DEF. It's also why every 1B has a negative DEF score (because even the best 1B defense doesn't give more value than average defense at shortstop) 

Max had a positive DRS in RF (+5) and right around average UZR (only slightly below zero, not enough to say below average).  DRS and UZR are the stats to look at when comparing Max to other RF (or any player to other players at the same position).

So no, Fangraphs DEF stat wasn't really hating Kepler's defense, it was adjusting for position.

All stats, even traditional ones (even offensive ones, even pitching ones), needs about three years of data to tell us what the players true talent level is.  To create a good 'mean'.  

Doesn't mean we toss out a players performance for one year and say it didn't happen, just means that the way he performed that one year, either offensively, defensively or hurling the ball, may not be what his true talent is.  One year defensive metrics can be used to gauge how a player did that year and defensive performance, just like offense and pitching, can fluctuate.

The worst thing Fangraphs did when explaining defensive metrics was explain the obvious (that obvious thing being that more data gives us a player's true talent level). They should have just reminded people that it's the same situation with all stats, not just defensive metrics. They do say one years worth of defensive data can tell us what the player did that year (particularly DRS, which is why I prefer to use that).  They talk about 1 year and 3 year samples as being stable, meaning one year to tell us what the player did for that year (yearly performance) and 3 years to get a mean (measure true talent level).

'The other thing to remember is that DRS isn’t going to work well in small sample sizes, especially a couple of months or less. Once you get to one and three-year samples, it’s a relatively solid metric but defensive itself is quite variable so you need a good amount of data for the metrics to become particularly useful.'

Right there it says one year.

Having said all that, defensive metrics will get better (but still better than eye test or errors/fielding %), just like better offensive metrics have been created (like wRC+) that are better than what we used to have.

SDI is bunk, unfortunately.

 

SDI adds the results from other defensive systems together -- heck just about every defensive system you can think of is taken and all added together to form SDI. 

 

Some systems count some things that others do not count, but there is considerable overlap. Any stat that has overlap (or is counted in multiple systems) will thus be counted multiple times by the SDI index. 

 

For example, with SDI, one error is suddenly counted several times inside of various algorithms that are trying to put forth a theory as to what an error is actually worth. So rather than SDI figuring out what an error is ACTUALLY worth (picking the best theory), SDI simply takes every definition of an error's worth and adds them all together. Suddenly an error's value is inflated into the stratosphere.

 

So if a player happens to be outside the norm on some trait that is counted in multiple systems, SDI explodes that value out. But what if a great, important trait is only counted in one or two systems, and the player excels at it? It's buried in the mess.

 

SDI might be the most comical baseball statistic ever and I'm a bit amazed that the SABR guys aren't embarrassed by its existence. 

Thanks - I have had major issue with this for some time. We were not good but this is a butch of junk.