Posted by sean on March 29, 2013
In a move to 1) reduce the number of negative WAR long-time players and 2) create a bit more consistency between our version of WAR and that on FanGraphs.com, we have agreed with them (their article about this change) to move the replacement level to .294 or 1,000 Wins for current seasons (30 teams * 162 games * (.500 - .294) = 1000). Pre-30 teams leagues will likewise be adjusted. Our previous Replacement Level was .320 or 875 wins, so we are now giving out 125 more wins than we were before. FanGraphs previously had a .260 replacement level, so they are going from 1,166 wins to 1,000. The net result is that these adjustments are given to players in a manner proportional to their playing time regardless of their quality of play.
The main outcome of this on Baseball-Reference.com is that everyone gets an increase in their WAR of about .2-.4 WAR/year for full-time guys, and over a long career this is a boost (4-5 wins).
The main benefit of this change is that Baseball-Reference.com and Fangraphs are now handing out WAR from the same sized basket, so any differences in WAR totals is now due to our differing methodologies for evaluating pitcher value or batter defense, etc rather than just an issue of scaling.
For example, the scales on our leaderboards are now very similar. Before, the ordering would have been similar, but the values would have been much different, now the ordering is similar and the values are similar.
All-Time Batting WAR (all #'s rounded)
B-R followed by Fangraphs,
#1, Babe Ruth-163, Babe Ruth-168
#2, Barry Bonds-163, Barry Bonds-164
#10, Eddie Collins-124, Stan Musial-127
#20, Joe Morgan-100, Nap Lajoie-101 (note we both now have exactly 20 100-WAR players)
#50, Reggie-74, Carew-73
#100, Lou Boudreau-63, Gary Sheffield-63
#500, Kevin McReynolds-30, Kevin Mitchell-30
All-Time Pitching WAR (all #'s rounded), Note that we have big methodology differences here, but the scale is the same.
#1, Cy Young-170,Roger Clemens-140
#2, Walter Johnson-153,Walter Johnson-126
#10, Phil Niekro-97,Pete Alexander-98 (note we both now have exactly 9 100-WAR pitchers)
#20, Nolan Ryan-84, Warren Spahn-81
#50, David Cone-62, Ted Lyons-62
#100, Hoyt Wilhelm-50, Mordecai Brown-48
#500, Denny McLain-21, Jim Scott-22
Another example to give you. This offseason B.J. Upton signed with the Braves and ESPN the Magazine noted that over the last three seasons FanGraphs gave Upton 13.9 WAR and B-R had him at 7.2 WAR a pretty major difference, but this difference was made up entirely of two things. 1) the different replacement level and 2) a difference of opinion on our defensive metrics.
Using the new unified replacement level, FanGraphs rates Upton (who they rate as an average fielder over the last three years) with 10.7 WAR from 2010-2012, while B-R (who ranks him as a poor fielder) has 7.3 WAR, however if you look at our oWAR stat, which treats every player as an average fielder, we now have Upton with 10.6 oWAR (compared to Fangraph's 10.7 oWAR). The same amount, so now the difference depends entirely on your defensive view of Upton.
Likewise on the career level, our differences for Jack Morris were reduced from 20 WAR to a smaller, but still significant 9 WAR.
Here is a rundown of articles either about this change or WAR in general.
- B-R: My Answer to I Don't Like How Complicated WAR Is and How It Is Constantly Changing
- FanGraphs:Unifying Replacement Level (their article about this change)
- FanGraphs: Why Replacement Level (on why replacement level is a useful concept)
- WAR replacement levels
- WAR Explained
- About with Full List of WAR Articles