Archive of SportingNews.com's old SOM Baseball forum

My article on how to evaluate the offensive and defensive contributions of a SOM card has been updated and greatly revised. The article can be found at:

http://silenceisdefeat.com/~meandean/som/OffensevsDefense.zip

I've also posted this notice to my blog.

Hope you find the article useful!

Thanks Dean!!
:D

Gracious of you to do this; looks like it's a lot of worthwhile information to process.....you can delete the other threads, though! :D

tws

This is incredibly useful information. For example, I had a platoon with Beuchle 3/e5, and Lind 1/e12. The question was, should i keep Beuchele or Lind in the late innings. Hands down, go for range over errors. Thanks!

Hey, dean, my Mac doesn't want to open this. It keeps saying it encountered a problem on the document. Any tips...I'm eager to see your updated work? Thanks!

BUMP!

This article (the Zip file in the link, PDF document) is an incredibly good article on how to evaluate run production mathematically!

bump again

Good work overall, Dean.

The main comment I have about your method is that I argue it only allows for comparisons of players who play essentially the same defensive position. You could still compare LF to RF (because 3e3 in LF has the same defensive impact as 3e3 in RF)... but otherwise, it's not valid to compare 3B to CF or 2B using the same ERP value. Here's why.

A certain amount of defensive imperfection is a part of the game. At each position, varying levels of competence are found. For instance: if you looked at all the players were principally 2B, then among those players, chose the top 30 by number of PA (equivalent of 1 per MLB team), the median defender in the group of 30 (i.e. the median defense of a typical starting MLB 2B) would be about a 2e16. If you went through the same exercise for 1B, you'd get about a 3e8.... and so on. In order to compare ERP across positions, I suggest you'd set these median defensive ratings as "par" for each position and adjust each player's rating accordingly.

Two examples:
(A) suppose prior to considering defense, a 2B's ERP is 60. Suppose he plays defense at 2e10 and the median starting-2B defense is 2e16. His par-adjusted rating would be [b:816242851f]63.6[/b:816242851f]: 60 - (-0.8) {for his own 2e10 defensive abilities} + (2.8) {the defensive runs allowed figure for a 2e16 2B [par at 2B]}.
(B) suppose prior to considering defense, a 1B's ERP is 60. Suppose he plays defense at 2e5 and the median starting-1B defense is 3e8. His par-adjusted rating would be [b:816242851f]64.8[/b:816242851f]: 60 - (1.0) + (5.8).

Under the unadjusted system, the 2B rating would be [i:816242851f]60.8[/i:816242851f] and the 1B rating would be [i:816242851f]59.0[/i:816242851f], leading us to believe that the 2B would have more value than the 1B.
Per the adjusted rating, the 1B is more valuable.

My argument, given equal raw offensive ability, the example 1B defensive ability relative to average available 1B is better than the example 2B ability in comparison to the average available 2B.

-----


One more-minor comment: If you do the math on the super-advanced fielding charts, they presume 666.66 PA per lineup slot (6000 PA per team) per year, instead of 648. The only impact to your method is that it slightly overstates the negative ERP effect of defensive errors, but since the E in ERP is "Estimated"... the overstatement is statistically insignificant.
[b:816242851f][/b:816242851f]

[quote:105a434093="zonachoke"]Per the adjusted rating, the 1B is more valuable.[/quote:105a434093]

The difficulty in comparisons like that is that there are many 1B who can hit like that, but very few 2B.

Looking at WAR, the 2B would be a far better choice than the 1B, right?

Yeah, basically if you adjust the defensive portion of the players for "average" to normalize them, you really should normalize the offensive portion of things also.