Since Rally has released his historic WAR values for the Retrosheet period (pitcher data from 1953 to current and hitter data from1955 to current) there has been many comparisons of players. The comparisons have been limited to only players that started their career after 1953/1955 due to incomplete data. I wanted to somewhat fill that void.
I took some batters and pitchers from Fangraphs and plotted there WAR vs Win Shares to get a regression equation to best estimate WAR when Win Shares is know. For Win Shares, I used this dataset provided by The Hardball Times. The system is not perfect by any means, but it will allow some comparisons of pre and post Retrosheet players. The following are the requirements and equations:
Requirements:
- Position WS = Fielding and Batting
- Pitching WS = Pitching
- Ignore pitcher batting unless it is the Babe as this will over punish the pitcher because they will have around 0 Win Shares hitting and this will lead to a negative WAR.
- I used only hitters with 300 at bats and 40 innings pitched. If players are below these values, they might have a low Win Share Total which would impact negatively the WAR values since low Win Share total would lead to a negative WAR, even if the batter didn't hit but a few times
Starting Pitcher (SP in Win Share download)
War = (WS - 3)/3
r-squared = 0.860
Relief Pitcher (RP in Win Share download)
WAR = (WS -1.8)/5
r-squared .81
Positional Players
(WS-9.3)/3 = WAR
r-squared = .8735
First, here a comparison of Bond's predicted vs actual WAR. His totals add up 174 WAR actual and 167 predicted WAR.
Here is an example of what they can be used for as in filing in the gaps in Bob Lemon's career for the time before 1954.
As always I am open to suggestions and comments.