WARP3’s systematic flaws with 19th Century players
I’m going to do an overhaul on pennants added this week (hopefully, if you want to volunteer for some data entry, let me know!), including revising for the WARP 2003 numbers (I’ll leave the old ones up, for reference), and adjusting for the overrating of fielding by setting the replacement level too low.
One thing I noticed, that troubles me from the Prospectus glossary:
Fielding Runs Above Replacement. The difference between an average player and a replacement player is determined by the number of plays that position is called on to make. That makes the value at each position variable over time. In the all-time adjustments, an average catcher is set to 39 runs above replacement per 162 games, first base to 12, second to 34, third to 26, short to 38, center field to 30, left and right to 20.” (emphasis mine)
This means that WARP3 systematically overrates 2B and underrates 3B from this time period. It also underrates 1B, who were at least as valuable as LF and RF defensively probably a little more valuable.
If I had the time, I could adjust this defensive spectrum, and re-rate the fielding component for each player, but that ain’t happening any time soon.
What I’d suggest is this: Using the same total of runs 39-38-34-30-26-20-20-12, I’d portion them this way, pre 1930:
39 C, 38 SS, 34 3B, 26 2B, 25 CF, 21 1B, 19 LF, 17 RF.
You could tweak that to your own specifications, but I think I’m being pretty conservative, you could easily make the case that 1B was equal in value to CF, though I won’t.
Doing that, you get an adjustment (using a 9.7 R/W converter) of:
3B: 34-26 = 8; 8/9.7 = +.82 W/season
2B: 26-34 = -8; -8/9.7 = -.82 W/season
CF: 25-30 = -5; -4/9.7 = -.52 W/season
1B: 21-12 = 9; 9/9.7 = +.93 W/season
LF: 19-20 = -1; 1/9.7 = -.10 W/season
RF: 17-20 = -3; -3/9.7 = -.31 W/season
I think that if you do this, you’re going to come out with rankings much more in line with the generally accepted historical rankings, especially with players like Jimmy Collins and Bid McPhee.
I’m not saying that WARP should be eliminated from the tool box when discussing 19th Century players. Just that it needs to be tweaked somewhat.
The mistake that Prospectus is making is that they think (implied by their formulas) that if Jimmy Collins were playing in 1990 he’d be a 3B. But most likely he’dve played 2B, because that’s where the more skilled fielder played after 1930. That’s the only way to explain how 3B outhit 2B after 1945, and 2B outhit 3B before 1920. From 1920-42 they were about even (who cares about 1943-45 :-)
Posted: December 02, 2003 at 04:53 PM | 61 comment(s)
Login to Bookmark