The following article is part of my ongoing look at the candidates on the BBWAA 2013 Hall of Fame ballot. For a detailed introduction to JAWS, please see here.
I've spent the better part of the past month examining the 2013 BBWAA Hall of Fame ballot in light of my JAWS system, the basis of which is comparing each candidate to the average enshrined player at his position in terms of Wins Above Replacement, and highlighting those candidates who are above that average, which I often refer to as "the standard" at the position. I've chosen to focus on that average (the mean, technically speaking) while fully recognizing that there exists a significant discrepancy in quality between players voted in by the BBWAA and those voted in by the various iterations of what we commonly refer to as the Veterans Committee.
When the Hall was founded in 1936, BBWAA voters were given jurisdiction over 20th century players, while a separate Old Timers Committee was given jurisdiction over 19th century players. Believe it or not, nobody stopped to think too long about how to handle players whose major league careers significantly straddled the turn of the century, and the resulting confusion left no 19th century player with the requisite 75 percent of the vote for the inaugural class, which is why Cy Young, whose career spanned from 1890 through 1911, wasn't voted in the first time around, while Ty Cobb, Walter Johnson, Christy Mathewson, Babe Ruth and Honus Wagner were; among that group, only Wagner (1897) reached the majors before 1900.
Eventually that was straightened out somewhat, but when the BBWAA shifted from an annual to a triennial balloting cycle, the backlog of candidates piled up high enough that it was tough for any player to gain the needed supermajority, so just one player was elected in 1942 and none in 1945. In an effort to compensate for this, the Old Timers Committee opened the floodgates for 20 players (and one manager) over the next two years. After adding two more players in 1949, the Old Timers Committee was disbanded, and not until 1953 did a replacement Veterans Committee come about.
Despite playing shorter schedules, with rules that were considerably in flux until just before the turn of the century, some of the players voted in by the Old Timers or Veterans Committees measure up well against those elected by the BBWAA — others, not so much. Over the years, the various iterations of the VC have skewed towards the latter, particularly when it comes to the offense-heavy era of the 1920s and 1930s, when .300 batting averages were a dime a dozen. Even so, in my work with JAWS, I've always felt that any voter (or otherwise interested party) reckoning with the definition of a Hall of Famer should consider the entire spread of enshrined players rather than just a subset, but I do think taking a peek at what a strict BBWAA standard at each position would look like can be revealing.
Before pulling back that curtain, I've got another wrinkle to add. Many smart individuals, fans of my system as well as critics, have questioned why I don't compare players to the median at the position rather than the mean. I've considered that option many times, and even written about it in years past. The basic problem is that so many more players exceed the median that even before this year's packed slate of 37 candidates, the average ballot yielded far more than 10 candidates worthy of election. That's an extreme position, to say the least, and one that would create a nightmare for any voter trying to apply it.
As an experiment, what I've decided to do is show the consequences of using these different cutoffs to construct a virtual ballot. I've designated any candidate who exceeds the median JAWS at his position a Level 1, any candidate who exceeds the mean JAWS (the current standard) a Level 2 and any candidate who exceeds the BBWAA average JAWS a Level 3. I'm throwing out all subjective considerations for this exercise — awards, milestones, postseason accomplishments, PED allegations — and also ignoring the career and peak subtotals lest anyone's eyes glaze over in the resulting flood of numbers. Given that, here's how the position-by-position standards look like at the various levels:
Note that for some positions such as first base, shortstop and third base, the three numbers are clustered fairly closely together, while for others such as centerfield and rightfield, they're spread quite far apart. At shortstop and third base, the median is actually higher than the mean. All of that is a reminder that within the small sample of Hall of Famers, WAR-based value isn't distributed evenly, either within a given position or across positions.
Taking another look at this year's 37-candidate slate, here's how the players on the ballot would classify:
Viewed through this lens, the current ballot has four players who exceed the strict BBWAA standard at their position, which is to say that their JAWS are above the positional averages of the players voted in by the writers since 1936: BArry Bonds, Roger Clemens, Jeff Bagwell and Mike Piazza. Four other players fall short of that but exceed the mean at their positions: Curt Schilling, Larry Walker, Alan Trammell and Tim Raines. All eight of those players made the cut on my virtual ballot.
Two other players exceed the median but not the mean at their positions: Kenny Lofton and Rafael Palmeiro, neither of whom made my ballot. Meanwhile, two players who did not exceed either the median or the mean at their positions did make my ballot: Craig Biggio and Edgar Martinez. In reviewing their candidacies, I argued that both should be measured against modified position standards reflecting their trajectories around the diamond — up-the-middle players for the former, since he spent significant chunks of his career at catcher and centerfield, and all Hall hitters for the latter, who spent most of his career as a DH out of injury concerns. By this alternative methodology, it's a bit harder to justify those votes.
Further down, it's worth noting that some of the ballot's more controversial candidates also fall below even the median at their position. That group includes both Mark McGwire and Sammy Sosa, whose candidacies are clouded by PED allegations, Jack Morris, who's well below any sabermetric standard for starting pitchers, and Dale Murphy, who's gotten a fair bit of attention given that this is his final year on the ballot.
Among this year's crop, the median produces a manageable 10-candidate slate, but in years past it didn't, and in future years that may be the case as well. If I were to shift to a system along those lines, it might mean advocating for the elections of a whole lot more players via either the BBWAA or Veterans Committee processes. Those ballots already appear quite full, and produce only a trickle of inductees as it is; widening the field would likely make it harder for any candidate to reach 75 percent.
Consider centerfield, for example: unenshrined players surpassing the WAR-based median include Ken Griffey Jr., Lofton, Andruw Jones, Carlos Beltran, Jim Edmonds, Jimmy Wynn and Willie Davis. Griffey and Edmonds aren't yet eligible for the BBWAA vote and Jones and Beltran are still active, while Wynn and Davis are eligible for nomination during the next Expansion Era ballot cycle in 2014; the last time around, in 2011, they didn't even make the final ballot, and they've fallen well short when appearing on previous VC ballots. Both are a significant distance below the mean, and if they were to join the ranks in Cooperstown, that number would drop by a small fraction, lowering the standards a hair. Incorporating some kind of striation into the JAWS system in the future is an idea I'm not entirely against, but it does have the potential of making the coming crowd of candidates even more unwieldy, and makes the system a bit harder to explain to someone exposed to it for the first time. JAWS has made significant inroads with actual voters over the years, and I'd hate to undo that by making the system harder to follow. So for now, I'm playing it straight down the middle, and sticking to the average as the standard by which candidates are measured.