war by position update...

In my previous post I did some analysis to highlight that individual positional WAR might be better at projecting wins then just singular team WAR as a sum of all those parts.

I noted:

NOTE: I do understand this could be refined even further to re-weight the players WAR values exactly correctly based upon their actual number of games at each position instead of the approach I took which was just to equally distribute those values.  Given the size of that specific sample and what type of change we’d be talking about, I would find it unlikely that would move the needle substantially here though. But I think it’s an interesting finding.

i.e. So if a 6 WAR player showed up as a C & 1b & DH in a single season, each position was credited with 2 WAR.

I went back to see if i could actually do this analysis in a reasonable way.  I was able to come very close.  Fangraphs has a handy defensive data link that actually was able to identify innings by position.

defense by innings

But this data only went back to 2002.  So i had to shrink my analysis dataset down a bit.

The other problem with this data was the DH.  Innings played at the DH position weren't tracked with regards to fangraphs fielding metrics. So i did have to make an assumption here.  For players who were DHs, I was able to see their total games on the season. I was also able to see any innings played through the year at any position.  So i multiplied the games played by 8 (for 8 innings per game), and then i subtracted out any innings spent at any position on the field and arrived at my "estimate" for DH innings played that season.

So by doing that to the DHs and using fangraphs data for all other positions i now had the total number of innings played by position for any player in a given year on a given team. I simply took each positions innings and divided it by the total innings to arrive at a percentage for each position, and then i applied that % to that players WAR.  So now a 6 WAR player who played 50% at C, and 25% at 1b and DH was given 3 C WAR and 1.5 for 1b and DH.

This is still a bit problematic, as the assumption here is now that performance is consistent across all fielding positions for a certain player (i.e. the stress of playing C doesn't for some reason make the guy a worse hitter).  I'm not sure how much of an overall impact that might have, and that may be very player specific (i.e. certain mental makeups can deal and others can't).

When i did this adjustment for positions, and taking into account data solely from 2002,

You can see this is slightly different then the values I posted before.  SS is actually no longer the most overvalued by WAR, it's 2b.  Maybe another reason why the Robinson Cano signing was bad?
With the DH making a surprising burst into actual value mode and RP is even more pronounced as being advantageous (though since the Pitching data was not impacted this gain is entirely driven by the time frame change to 2002+ from 1985+).

So that RP point made me think, if the RP change was that pronounced soley due to the time period, was my original approach actually that far off? Or could most of these shifts be explained really just by 2002 to 2013 vs the original 1985 to 2013.  So i went back to my original approach and re-ran the numbers just changing the time period to now only be 2002 to 2013.

Sorry it's not in the same order, but you can see that it's basically the same.  The vast majority of the delta appears to be driven by the time frame change, and not the reallocation of positional value to a more precise percentage based on innings versus just equal weighting positions. I've lined up the outputs below to compare the values.

Given similarity i'd say that the extra effort required to do the precise positional allocation is likely not worth the effort. But if you've got it great.

The better question as a result of this updated analysis is whether or not you think it's more appropriate to use the smaller (2002+) or larger (1985+) dataset in making these determinations.

I again also recognize it's possible a player just isn't as good a hitter when at a specific position, and this could result in changes, but given the above analysis already showed the changes to be minor, i would expect those changes to be even less impactful, which is to say hardly at all.