An Exploration: Taking WAR Out of Hall RatingJan 1, 2017 by Adam Darowski
The Hall Rating formula has remained mostly untouched since the initial launch of the Hall of Stats in 2012. At its core, Hall Rating is a combination of Baseball-Reference’s Wins Above Replacement (WAR) and Wins Above Average (WAA) run through a series of adjustments. Lately, I’ve been having impure thoughts of ditching the WAR part of it and focusing on WAA.
There are a few reasons. First, while WAR is used more often than WAA in contemporary baseball analysis, I don’t believe WAR is the better metric for measuring a Hall of Fame case. My hypothesis is that WAA by itself (with adjustments) could do a better job of capturing the value of a player’s Hall of Fame case. WAR is incredibly useful for quantifying the value of a player in terms of roster construction, but that’s not what we’re doing here. We’re looking for players who stood out above their entire peer group. Rather than giving extra credit to the value generated above league average, maybe we should consider only that?
Second, some of the adjustments I make in Hall Rating are applied to the WAR component and some are applied to the WAA component (further confusing things, some are applied to both). I’d feel a bit better if they were all applied to the same component.
Lastly (and least importantly), while there are already many, many differences between Hall Rating and Jay Jaffe’s excellent JAWS framework, focusing on WAA while JAWS focuses on WAR would be a bit of a differentiator.
This article contains some in-progress tests of how Hall Rating would look if we removed the WAR component. It’s a little complicated because the adjustments I made to WAR now need to be made to WAA. Here is a rundown of those adjustments:
- I start with separate components for position player WAA and pitcher WAA.
- I apply the catching adjustment to position player WAA.
- I apply the relief pitching adjustment to pitcher WAA.
- I apply the season length adjustment to position player WAA.
- I apply the season length adjustment to pitcher WAA, but only for strike- or war-shortened seasons.
- I apply the 19th century pitching adjustment to pitcher WAA for seasons before 1893.
- After the adjustments, if either the position player or pitching WAA component is negative, I simply drop it (as a way to measure only peak seasons).
- I then combine the two components to make adjWAA.
Of course, this test is far from perfect so far. I’ve run into a few issues:
19th Century Hitters
In Hall Rating’s current formula, the schedule adjustment is only applied to the WAR component and not the WAA component. Adding it to the WAA component gives it a much bigger effect.
- Cap Anson: From 215 to 245 (+30.1)
- Ross Barnes: From 82 to 121 (+39.0)
- Deacon White: From 114 to 129 (+15.4)
Good hitting pitchers (and the Union Association)
In the current formula, I count pitchers’ hitting WAR but not pitchers’ hitting WAA. Since replacement level and average level are the same in this case, it seemed like too much of a boost to double-count it.
The same is true for the 1884 Union Association. This was a weak league, so I added this adjustment to knock these players down a bit.
In this initial test, I’m only using WAA so in essence these are double counted again (not really, but I’m only using one metric and it is included in that one).
- Carlos Zambrano: From 84 to 101 (+16.5)
- Wes Ferrell: From 111 to 139 (+28.3)
- Jack Glasscock: From 131 to 150 (+18.9)
- Fred Dunlap: From 72 to 100 (+27.7)
These adjustments help two of my personal favorites—Ferrell and Glasscock. I believe these two are absolutely Hall-worthy, but Hall Ratings of 140 and 150 seem a bit much.
Also, how about that Carlos Zambrano?
Great-hitting 19th Century pitchers see benefits from both of these issues.
- Bob Caruthers: From 96 to 115 (+19.6)
- Al Spalding: From 92 to 115 (+22.1)
The Current Ballot
These are the effects on the key players on this year’s BBWAA ballot (reflected in Hall Rating change):
- Barry Bonds: +35.1
- Roger Clemens: +16.8
- Curt Schilling: +11.5
- Jeff Bagwell: +5.9
- Mike Mussina: +0.7
- Ivan Rodriguez: -8.2
- Larry Walker: +6.0
- Edgar Martinez: -1.1
- Manny Ramirez: -8.1
- Tim Raines: -8.9
- Sammy Sosa: -0.1
- Gary Sheffield: -4.3
- Vladimir Guerrero: -6.6
- Jeff Kent: -7.0
- Fred McGriff: -9.1
- Jorge Posada: -10.5
- Billy Wagner: +0.2
- Trevor Hoffman: -1.8
- Lee Smith: -4.3
In most cases, the higher the Hall Rating of the player, the bigger their boost is. Schilling and Mussina, who currently have similar Hall Ratings, are interesting. Schilling sees a bigger boost as he’s more of a peak candidate. Mussina was also dominant, but was more consistent for a longer period of time.
This adjustment isn’t particularly kind to Tim Raines, knocking him down to a 119 Hall Rating. That’s still the median Hall Rating for a Hall of Famer, though.
These players see the biggest drops. Most of these aren’t terribly surprising—they are players who were solidly average (or slightly above) for a long period of time.
- Sam Rice: -21.8
- Harry Hooper: -21.3
- Jimmy Dykes: -21.1
- Joe Judge: -20.3
- Johnny Damon: -19.9
- Omar Vizquel: -19.8
- Clyde Milan: -19.3
- B.J. Surhoff: -19.2
- Wally Pipp: -18.8
- Ossie Bluege: -18.8
This adjustment knocks Omar Vizquel below a 50 Hall Rating. Some people insist he is a viable candidate, but I just don’t see it.
So, where do I go from here? Obviously this isn’t fine as is. My to-do list:
- Handle pitchers’ hitting stats better
- Tweak the schedule adjustment for 19th century hitters
- Apply an adjustment for the Union Association (and perhaps the Federal League, which was also pretty weak)
- Decide if this is really something I want to do
Whether I opt to go this route or not, it has been an interesting project.
Try it out for yourself. I’ve included all the before-and-after numbers in a Google Spreadsheet. Find anything interesting? Please share in the comments.