Projecting the 2017 Hall of Fame vote: Who's on track to get into Cooperstown?
- With just one week left until the announcement of the 2017 Hall of Fame class, let's take a look at the ballots we know so far to see which candidates are sitting pretty for election and which have more narrow hopes.
Due to scheduling issues, this year's Hall of Fame election cycle features the longest delay between the BBWAA ballot submission deadline (Dec. 31) and the announcement of the results (Jan. 18) in recent memory. With one week to go before the big reveal—which will be aired on MLB Network at 6 p.m. ET next Wednesday—it's worth a peek at the potentially historic possibilities.
Via Ryan Thibodaux's indispensable Ballot Tracker, as of Wednesday, 185 known BBWAA voters and six anonymous ones—a contingent that's estimated to be around 44% of the electorate—have published or otherwise shared their ballots. Among that group, three candidates have received more than the necessary 75% of the vote, putting them on track for election: Tim Raines (91.6%), Jeff Bagwell (91.1) and Ivan Rodriguez (80.1). What's more, both Vladimir Guerrero (74.3%) and Trevor Hoffman (73.3) are very close to the cutoff as well, leaving the door ajar for a five-man class.
The writers have produced exactly one such class: the 1936 inaugural Hall of Fame group of Ty Cobb, Babe Ruth, Walter Johnson, Christy Mathewson and Honus Wagner. Even a four-man class would be historic, as there have only been three so far: 1947 (Mickey Cochrane, Frankie Frisch, Lefty Grove and Carl Hubbell), '55 (Joe DiMaggio, Gabby Hartnett, Ted Lyons and Dazzy Vance) and 2015 (Craig Biggio, Randy Johnson, Pedro Martinez and John Smoltz). Three-man classes are also rare, as there has only been one in this century (2014, with Tom Glavine, Greg Maddux and Frank Thomas), just five since the writers returned to annual voting in 1966 ('72, '84, '91 and '99 being the others) and eight overall ('37, '39, and '54 from the years when the length of eligibility windows and the time between election cycles was in considerable flux).
It's important to note that the ballots in the Tracker are a non-random sample: They tend to come from younger and more analytically inclined writers, but that generalization should be understood in relative terms. A larger share of the private ballots than the public ones are understood to come from writers who are no longer working for BBWAA-affiliated publications but haven't yet been purged from the rolls for being 10 years removed from regular coverage (the result of a rule change from last year); they may no longer have outlets to publish their ballots and may not be as inclined toward transparency as active writers.
As a group, the private voters are definitely less generous when it comes to checking off names. For the 2016 cycle, when 70.5% of the 440 ballots were published either before or after the results were announced, the public ballots contained an average of 8.23 votes, but the private ones just 7.28. For 2015, when 60.3% of ballots were eventually published, the split was 8.73 to 7.95; that year's overall average of 8.42 names per ballot broke the previous year's record (8.39). For that year, the first that Thibodaux ran the Tracker and the first year that a majority of ballots were published (52.9%), the split was 8.61 to 8.19.
Candidates favored by analytics (such as Raines, Bagwell, Edgar Martinez, Mike Mussina and Curt Schilling) have generally received far greater public support—often by more than 10 percentage points—than private. Likewise for those linked to performance-enhancing drugs, even without a positive test (Bagwell, Barry Bonds, Roger Clemens, Gary Sheffield and the recently elected Mike Piazza for example). Meanwhile, candidates whose cases are driven more by traditional statistics (Hoffman, Fred McGriff, Lee Smith and Billy Wagner) have tended to receive greater support from the private ballots. Those trends are worth bearing in mind with regards to Rodriguez (whom Jose Canseco claimed to have injected with PEDs in his 2005 book, Juiced), Guerrero (who is about eight points below the JAWS standard among rightfielders but owns a .318 lifetime batting average, an omission that would be unprecedented) and Hoffman (who retired as the career saves leader); their shares of the vote could converge toward 75%, and it's a tossup as to which side of the line they wind up on.
As you'd expect from an increasingly Hall-savvy public, several people have attempted to predict the outcome by taking such factors into consideration. Nathaniel Rakich (@baseballot), who published the most sophisticated explanation for how his system works at The Hardball Times last week, uses a model that averages the public-private differential over the past three elections to calculate an adjustment factor for each candidate. For those with only one or two years of data, he averages what's available. For first-year candidates, he "finds 'veteran' candidates with whom the rookies’ votes are well correlated and adjusts their exit polls proportionally." For example, ballots that contain Manny Ramirez tend to include Bonds and Clemens, so his adjustment factor is based on their public-private differential. For Rodriguez, he has opted to use Piazza (who was elected last year despite having previously been linked to PEDs) as an analogue.
Rakich's model predicted every candidate within 3.5 percentage points last year, with an average error of 1.5 points, but those are based on the final pre-election public numbers as opposed to intermediate ones. When I asked him about this last week, when roughly 160 ballots had been published, he responded that relative to that juncture in 2016, he was off by around four points on Bagwell and by five on Smith and Mark McGwire. In the end Bagwell was the biggest outlier; in Rakich's final projection, which was done with 48.4% of the ballots published, he predicted the first baseman would receive 75.1% of the vote, but he finished at 71.6.
All of which is to say that his system isn't foolproof, but it is worth noting. Here's how things stack up at the 190-ballot mark (one fewer than in the Ballot Tracker).
|player||% public||adjustment||Est. % private||Proj. final|
As you can see, Rakich is projecting Hoffman and Rodriguez will both to get in, with Guerrero barely missing. What's more, he's projecting nine candidates to break 50%, something that has only happened once before, in 1947. While there have been several elections featuring seven candidates with at least 50%, only in 1951 were there eight. What's even more interesting is that none of the four other candidates projected to cross 50% has done so before. Historically speaking, every candidate who has reached that level has been elected eventually, either by the writers or a small committee, save for Gil Hodges, Jack Morris and current candidates Smith, Bagwell, Raines, Hoffman and Schilling—a group whose numbers appear poised to dwindle.
Bonds and Clemens, each in their fifth year of eligibility, topped out at 45.2% and 44.3% last year, respectively, but they're surging thanks in large part to Bud Selig's recent election and/or a general softening of attitudes when it comes to PED-related candidates, as discussed previously. Martinez and Mussina, the pair that gained the most ground last year, continue to make significant strides toward election. Martinez, now in his eighth year, gained 16.4% from 2015 to '16 to reach 43.4% and is setting himself up for a Raines-like ascension over the final two years of his candidacy; his net gain of 30 votes from returning voters thus far is the most of any candidate. Mussina, now in his fourth year, gained 18.4%—the most of anyone on the ballot—to reach 43.0% and could see another double-digit gain this time.
Schilling, meanwhile, vaulted from 39.2% in 2015 to 52.3% last year to break 50% for the first time, but he's projected to fall back below that line, as 23 writers who previously supported him have omitted him from this year's ballot, largely due to his inflammatory comments on social media (he's also gained 13 votes, for a net of -10). It's worth noting that Mussina doesn't appear to be gaining much direct benefit from Schilling's meltdown; the Tracker shows only two voters who have added the former and omitted the latter relative to their 2016 ballots.
Back to the top of the ballot: In a methodology separate from Rakich’s, Jason Sardell (@sarsdell) estimated that Rodriguez has a 63% chance of election, Guerrero 55% and Hoffman 50%. His model is based on Bayesian probability. As he explained, his public-private adjustment factor is based on categorizing 2016 voters “into a general category (full ballot, anti-Bonds, etc.) and treat[ing] 2017 data as exit polls.” For example, Hoffman has gained 54% among last year’s 81 full-ballot voters who didn’t include him. By estimating how many more voters fit that category this year, Sardell can simulate the probability of his getting elected.
Given the possibility of a near miss, it’s worth noting that Pie Traynor (1947), Nellie Fox ('85) and Biggio (2014) all missed by two votes, but nobody has ever missed by one; both Biggio and Traynor were elected the next year, and Fox (who had already passed away at that point) wasn’t elected until '97. On a more positive note, a three-man class would run the total of BBWAA-elected candidates to 12 over the last four years, matching the total from 1936 through '39 as the highest over a four-cycle span. Even if it's just Bagwell and Raines who get elected—and things are looking very good for both, as each has already picked up more new votes than the total margin by which they missed last year—the total of 11 candidates elected over the past four cycles would match the overlapping 1952 to '55 and '53 to '56 runs for the most since the '30s.
A four- or five-man class would produce history on this front as well as on its own accord, though anyone who thinks such a result will solve the ballot's backlog in one fell swoop should remember that next year's ballot will introduce the JAWS-approved Chipper Jones, Jim Thome and Scott Rolen, as well as Andruw Jones, who is slightly below the standard but still ranks 10th among centerfielders.
The next several days could provide a reality check on the current optimism regarding a large class. Both ESPN and MLB.com voters tend to publish their ballots just before the election, and as a group, they tend to include fewer names per ballot than the other public voters, meaning that the current rate of 8.66 names per public ballot will come down. Still, there’s obviously a lot of suspense ahead. Stay tuned.