Who’s On Track To Make The Baseball Hall Of Fame?

You may have heard there’s an election coming up. It’s expected to be close. And given that it’s conducted among a small and homogenous group of voters, it’s very hard to predict the results. Yes, on Tuesday, the results of the 2020 election for the Baseball Hall of Fame will be revealed. What did you think we were talking about?

Every winter, eligible members¹ of the Baseball Writers’ Association of America vote on the players who will be honored with a spot in Cooperstown the following summer. Players need to receive at least 75 percent of the vote to be elected, and they are allowed to stay on the ballot for up to 10 years. But if a player hasn’t been elected after a decade’s worth of attempts — or if he falls below 5 percent of the vote in any one election — he won’t appear on subsequent ballots.²

So who’s on track to get elected, and who’s in danger of lapsing off the ballot? Thanks to the work of Ryan Thibodaux, we don’t have to guess. Many Hall of Fame voters share images of their ballots or write columns explaining their votes before the full results are announced, and Thibodaux and his team of volunteers tally up these partial results in their leviathan Baseball Hall of Fame Ballot Tracker. So far, they’ve hunted down 165 of the 412 ballots expected to be cast this year, and the results appear promising for several candidates:

New York Yankees great Derek Jeter is currently polling at 100 percent in his first-ever³ Hall of Fame election, giving hope that he could be the second player ever (after his longtime teammate Mariano Rivera just last year) to be elected unanimously. Larry Walker and Curt Schilling have also appeared on more than 75 percent of the ballots released so far. That’s especially important for Walker, who is in his 10th year of eligibility; this is his last chance to be elected through traditional means.⁴

But just as importantly, the Tracker shows other candidates — who as recently as last year had only token Hall of Fame support — gaining momentum for possible election in future years. A whopping seven players (Scott Rolen, Gary Sheffield, Billy Wagner, Walker, Todd Helton, Andruw Jones and Jeff Kent) have flipped, on net, at least 24 “no” votes from last year into “yeses.” This is likely because, with the election of 11 players in the last three years, the recent logjam of Hall-worthy players on the ballot is finally clearing up. (Voters are allowed to vote for up to 10 players at a time, and many voters used all 10 of their votes the last few years — but now, those slots are up for grabs again.)

But be careful — the Tracker’s numbers will mislead you if you take them literally. In reality, most of these players will finish significantly lower in the final results than they sit in the Tracker. That’s because the sample of this “poll” is not random: Voters who make their ballots public ahead of time also share other characteristics in common, like voting for more candidates rather than fewer. And voters who choose to stay private collectively appear to vote more based on traditional, rather than advanced, metrics and take a firm stand against steroid use, which means they are especially unlikely to vote for candidates like Rolen (a stathead favorite) or Barry Bonds (an alleged steroid user).

Someone who’s interested in, say, election modeling can use these patterns to predict the final Hall of Fame election results based on the Tracker’s data — and that’s exactly what I’ve done. With 165 ballots made public as of 11 a.m. on Jan. 17, the below would be my 2020 Hall of Fame projections if the election were today and no more ballots got released. (Of course, more ballots will almost certainly drop over the weekend, making these projections merely preliminary; as more ballots are made public, the accuracy of my forecast rises. Follow me on Twitter for real-time updates, including a final projection on Tuesday afternoon just before results are announced.)

Here’s the methodology in a nutshell: For every player returning to the ballot, I calculate an “adjustment factor” based on how big of a dropoff the player has historically suffered from public to private balloting, then subtract this adjustment factor from his current public vote share to come up with his estimated private vote share. In a new twist this year, the adjustment factors also take into account how close a player is to being elected; it turns out that the more public support a player has, the steeper his dropoff on private ballots — until his public support gets above 75 percent, at which point the dropoff abates again (the bandwagon effect, perhaps?).⁵ For players appearing on the ballot for the first time, there is no past dropoff data to go off, so I estimate their private vote shares based on the predicted private vote share of players with whom their votes are highly correlated.⁶ Finally, players who have received fewer than two public votes just get an adjustment factor of zero; it’s pointless to split hairs over whether Paul Konerko will get one vote or two in the end.

Contrary to the Tracker (which isn’t meant to be a predictor), my system expects that Jeter will be the only player elected on Tuesday. He is currently projected to receive 100 percent of the vote, although this is more art than science; it would take only one stingy private voter to rob Jeter of unanimity.⁷ Idiosyncratic behavior like that is virtually impossible to predict, but practically speaking, there is no difference between Jeter getting 100 percent or 99.8 percent; he will easily join the Hall of Fame’s class of 2020. Meanwhile, I expect Schilling to perform around 20 points worse on private ballots than on public ones; his numbers on public ballots simply aren’t good enough to withstand that kind of hit, so it’s a very good bet that he will not get elected this year.

The only real suspense is around Walker. Currently, I project him to fall just short, with 73.3 percent support. But as someone who believes Walker belongs in the Hall of Fame, I’m comforted by the fact that models like mine come with plenty of uncertainty. In an eerily similar situation at this time last year, I thought Mike Mussina would get just 73.0 percent of the vote — but he wound up getting elected with 76.7 percent. Small errors like that happen all the time with my model; it’s just that they don’t normally mean the difference between election and elimination.

So I’m not writing off Walker just yet. I’m not the only Hall of Fame prognosticator in town, nor am I the best; for the last two years, that honor has gone to Jason Sardell, whose model divides voters into lanes⁸ and then projects how many votes each player will gain over last year based on how many net votes each player has flipped from “no” to “yes” in each lane so far. (In a treat FiveThirtyEight readers are sure to enjoy, Sardell uses a Bayesian model that spits out not only a predicted vote share, but also each player’s probability of being elected.) And as of Jan. 8, Sardell was giving Walker a 76 percent chance of achieving baseball immortality. This one is going to come down to the wire.