from Amazon
Monday, November 03, 2008
My 1B is better than your 1B
Chris Dial offers his OPD metric (off, def, pos, no baserunning). Since Google Docs (and Edit Grid) are now blocked at the office, I can’t make a more informed opinion on the results. So, I will just respond to the comments of the article. The first is from Chris Dial:
If the Cubs had gotten Joey Votto’s performance instead of DLee’s, they would have been a better team. Players are always (IMO) properly compared to how the position they played hits.
Now, in my opinion, we have pools of position: Catcher, Infielder, Outfielder, Firstbasemen/DH. The Infielder pool is all current 2B, SS, 3B, plus catchers. The OF pool is all current OF, plus IF and C. If you are talking about comparing Russell Martin to Brian McCann, then fine. But, the pool of Firstbasemen/DH is every player in MLB. 1B is not a “position” that is so fixed that you are limited to just comparing to 30 or 50 players. This is why I reject, out of hand, any overall player lists that try to compare Ryan Howard to Jimmy Rollins, without acknowledging the fact that the average 1B does not necessarily have the same impact as the average SS.
Now, HOW to make the translation is subject to lots of disagreement, and I’ve bounced back and forth alot on the issue. At the very least, we should use a long-term offensive replacement level to make the comparison (so if the average 1B over the last 10 years is +12 runs and the SS is -9 runs, then the translation is 21 runs). My preferred route is to look at players who play at multiple positions. To that end, I use +7.5 for SS and -12.5 for 1B. That said, Chris gives us the numbers, so we can manipulate it as we need.
If Chris is using the original XR equation, then he overvalues singles and undervalues doubles. The gap between the two is not .21 or .22 runs or whatever XR purports. Given that we have actually PBP data that tells us the gap is close to .30 runs, why not use the better numbers?
Ron Johnson says:
You’ll get a good sense of where replacement level is by looking at the team OPD. My sense is that replacement level is something close to -30 per Ripkenseanson (IE every inning of every game)—maybe -35, and yes some teams did worse
Ron, of who Chris is a huge fan, so we should give him more deference on that basis, seems to look at unregressed sample performance numbers. This is wrong. The replacement level that I use is 2.25 wins per 162G. You can argue for something a bit more, but not 3 wins. That’s extremely high. 3 wins per 162 G means that replacement-level nonpitchers are 27 wins below average. I’m not sure where he has pitchers, but let’s say 20 wins below average. So, his replacement-level team would be some 47 wins below the average of 81 wins, or 34 wins (or a .210 team). That I’m not buying unless I see a real good selling job.
There was also discussion on JJ Hardy. As is my basic rule, ignore every comment from any single person, as they are all equally worthless. Mine included. What I do have, from an observational standpoint, is 24 hardcore Brewers fans (you know, the people who actually see ALOT of him) who rank him as 11th out of the 30 regular shortstops, making him a bit above average. The Fans see him as one of the slowest SS around, but otherwise, pretty solid otherwise. (He’d make a bit better 3B.)