The Gate Could Be Closing on Future Hall of Fame Era Committee Inductees by Jay Jaffe July 20, 2022 Democrat and Chronicle This weekend in Cooperstown, six Era Committee candidates will be inducted alongside the BBWAA-elected David Ortiz. Among them are some of the most long-awaited honorees whose supporters agonized for decades over their being shut out, both before and after their deaths. Negro Leagues player/manager/scout/coach/ambassador Buck O’Neil and Negro Leagues and American League star Minnie Miñoso both hung on well into their 90s hoping they could see the day of their induction but died before it happened. Star first baseman and manager Gil Hodges died of a heart attack at age 47, before his candidacy became the ultimate “close-but-no-cigar” example, both via the BBWAA and Veterans Committee processes. Black baseball pioneer Bud Fowler, who was raised in Cooperstown, went largely unrecognized until the centennial of his death in 2013. Tony Oliva and Jim Kaat, both of whom are 84, are thankfully alive to experience the honor, but they, too, had a long wait, after falling one and two votes short, respectively, on the 2015 Golden Era ballot. The festivities will be tinged with more than a hint of bittersweetness due to the deferred honors, but there won’t be any shortage of joy and catharsis that these men are finally being recognized. Yet even as they take place, it feels as though a gate is swinging shut behind them — one that may not open again for awhile given the the shakeup of the Era Committee process that the Hall announced in April which reduced the numbers of committees, candidates, and votes available. I won’t rehash the road to this point (you can see the gory details in the aforementioned link), but here’s the new format, which will roll out in this order over the next three years starting in December: December 2022 (for Class of 2023): Contemporary Baseball – Players. For those who made their greatest impact upon the game from 1980 onward and have aged off the BBWAA ballot. December 2023 (for Class of 2024): Contemporary Baseball – Managers, Umpires, and Executives. For those who made their greatest impact upon the game from 1980 to the present day. December 2024 (for Class of 2025): Classic Baseball. For those who made their greatest impact upon the game before 1980, including Negro Leagues and pre-Negro Leagues Black players The Classic Baseball Era Committee now has purview over all of the candidates previously covered by the Early Baseball (1871–1949) and Golden Days (1950–69) committees — the two that produced this weekend’s honorees and which otherwise weren’t scheduled to convene again for 10 and five years, respectively — as well as about half of those covered by the Modern Baseball (1970–87) one. In other words, voters for that ballot now have to weigh candidates whose contributions may have taken place over a century apart. What’s more, where there were 10 candidates apiece for each of those ballots under the older system, the new ones contain only eight, and where the 16 committee members (a mixture of Hall of Famers, executives, and writers/historians) could previously vote for four of those 10 candidates, that number has been reduced to three. Candidates will still need to receive a minimum of 75% of votes to be elected. In other words, there’s a new bottleneck in place for the older candidates, and it has happened just as the Negro Leagues and pre-Negro Leagues candidates — players and non-players alike — finally returned to eligibility after the books were closed on that period following the aforementioned 2006 election, which produced 17 honorees but froze out O’Neil. For those who make it to the ballot, the math that was already very tough is undeniably tougher. Instead of a maximum of 64 votes spread across 10 candidates (an average of 6.4 per candidate), there are now 48 spread across eight candidates (six per candidate). Electing four candidates from a single slate, which happened for the first time on the 2022 Golden Days ballot, would require each of those four to receive exactly 12 votes. About that math: When I wrote about the changes in April, I linked to a December 2014 piece by Joe Posnanski, written on the occasion of the voters’ shutout of the Golden Era Committee. That decision slammed the door in the faces of Miñoso and Dick Allen one final time before each passed away; also falling short on that ballot were Kaat, Oliva, and Hodges. All but Allen were elected in December (he missed by one vote!), and with the exception of Miñoso, whose short but stellar Negro Leagues stats are now included in his major league career totals, it’s not like their credentials improved. Anyway, in that piece, Posnanski brought up a name that’s particularly familiar around these parts: Back to the math. Tom Tango explains it this way: Let’s say all ten candidates on the ballot were equally qualified for the Hall of Fame. That’s not quite true here, but it’s a good starting point — you had 10 good candidates. If they’re all equally good candidates, then each one had a 40% chance of getting picked for a ballot — 10 players on the ballot, voter chooses four, 40% chance. Pretty simple. Well, if a player has a 40% chance of being on one ballot, his chances on making 12 of 16 is … get ready for it, less than 0.5%. That’s not 5% — it is less than one-half of one-percent. 995 times out of a 1,000, the player would NOT get elected. And remember, that’s assuming every voter uses all four of his votes. In light of this, I asked Dan Szymborski, our staff probability expert, if he could help me figure out a way to illustrate the impact of the changes. He rose to the occasion by creating a Monte Carlo simulation, a model along the lines of what we use to calculate our Playoff Odds. In this case, he ran one million simulations for various scenarios. For starters, using Tango’s simple estimate from randomness, when all 10 candidates are equal and thus have an equal chance (40%) of being named on a four-slot ballot, the yield is an average of 0.049 inductees per year, and the committee elects no one in 95.2% of the simulations. Lower that to eight candidates, with each candidate having a 37.5% chance of being named on a three-slot ballot, and the yield drops to 0.021 inductees per year — basically from one every 20 years to one every 48 years — with nobody elected 97.7% of the time. We know that not all candidates are created equal; some have better numbers and more impressive accomplishments than others and are more likely to capture the voters’ attention. It’s easy to look at an Era Committee ballot and identify a few candidates who are basically ballast — guys who have been up for election several times before but have never or rarely gotten close and are likely to be overshadowed. On the other end of the spectrum are guys who might be in their first appearance in this format and scan as the most likely honorees, particularly if they have an easy hook. Think Jack Morris with his 254 wins, Harold Baines with his 2,866 hits, or Fred McGriff with his 493 homers. Suppose, for example, that we take 10 candidates of varying strengths and thus various probabilities of being named on an individual ballot. In this scenario, the best candidate has a 72% chance of appearing on a single ballot. That doesn’t mean that he’s going to get 72% of the vote every time, but that he’ll receive an average of 0.72 x 16 = 11.52 votes per simulation, sometimes more — enough to be elected — and sometimes less. Each of the nine other candidates has odds of appearing that are about 7% less than the candidate above him in the rankings, thus accounting for all 64 possible voting slots: 10-Candidate Model, 4 Votes Per Ballot Candidate Rk Ballot Odds Votes Per Sim 1 72% 11.5 2 65% 10.4 3 58% 9.3 4 51% 8.2 5 44% 7.0 6 36% 5.8 7 29% 4.6 8 22% 3.5 9 15% 2.4 10 8% 1.3 Total 64.0 Running those odds through Dan’s Monte Carlo simulation, this scenario produces an average of 0.989 inductees per year, with a shutout 28.2% of the time. Now, if you then take the top eight of those candidates and reduce their shares proportionally to account for the fewer voting slots (48, via 16 voters with three slots apiece), the top odds start around 57% like so: 8-Candidate Model, 3 Votes Per Ballot Candidate Rk Ballot Odds Votes Per Sim 1 57.4% 9.2 2 51.7% 8.3 3 46.0% 7.4 4 40.4% 6.5 5 34.7% 5.6 6 29.0% 4.6 7 23.4% 3.7 8 17.7% 2.8 Total 48.0 That scenario reduces the average number of inductees per year from 0.989 to 0.195, with a shutout 81.5% of the time. Eep! However, if the top player on the first ballot remains the favorite and still has 72% odds of appearing on a single ballot because his candidacy is so strong, with the shares of the others votes each reduced… 8-Candidate Model, 3 Votes Per Ballot Candidate Rk Ballot Odds Votes Per Sim 1 72.0% 11.5 2 62.0% 9.9 3 52.0% 8.3 4 42.0% 6.7 5 33.0% 5.3 6 23.0% 3.7 7 13.0% 2.1 8 3.0% 0.5 Total 48.0 …then the drop is only to 0.801 inductees per year, with a 35.2% chance of a shutout. In other words, the changes in the number of candidates and the number of votes per ballot reduce the yield by a lot, but the amount of the reduction depends on the individual players; a candidate who stands head and shoulders among the rest makes the election likely to produce an honoree. So what happens if, like the 2022 Golden Days ballot, a subset of candidates are clearly favored ahead of the rest? In that particularly case, the top five vote-getters (Miñoso with 14, Hodges, Kaat, and Oliva each with 12, and Allen with 11) accounted for 61 of the possible 64 votes, with at most three distributed among the other five candidates (their shares were reported as “three or fewer votes”). If you have five out of 10 candidates who have a 70% chance of appearing on a ballot, and the other five with a 10% chance (accounting for all 64 votes), the yield is 2.25 inductees per year, with a shutout just 5% of the time. If you have a similar split on an eight-candidate ballot, with four candidates having 70% chances and four with 5% chances, you get 1.8 inductees per year, with a shutout 9.2% of the time. As you’d expect, it’s the agreement among the voters — the consensus coalescing around a smaller subset of candidates — that’s the largest factor in determining the yield. When I spoke to Hall president Josh Rawich about the changes in April, he conceded that the new format makes it “more challenging to get on a ballot.” I suggested something to the effect that it would reduce the number of reheated candidacies (my term, not his, it should be pointed out). “There was definitely a feeling [among the Hall’s board members] that we wanted to make sure that we’re not looking at a lot of the same players every single time,” he replied. “Once somebody’s had a chance to be reviewed a number of times, it’s time to let somebody else get looked at.” The problem, to relate it to the modeling above, is that if you’re attempting to get rid of the candidates with the 5% or 10% odds, you’re going to create either a new crop of those same types because somebody will inevitably trickle to the bottom, or a flatter distribution of the odds and therefore a much lower yield. Here are two more eight-candidate scenarios: 8-Candidate Model, 3 Votes Per Ballot, Flatter Distributions Candidate Rk Ballot Odds Votes Per Sim Ballot Odds Votes Per Sim 1 60.0% 9.6 50.0% 8.0 2 60.0% 9.6 50.0% 8.0 3 50.0% 8.0 50.0% 8.0 4 40.0% 6.4 50.0% 8.0 5 25.0% 4.0 25.0% 4.0 6 25.0% 4.0 25.0% 4.0 7 20.0% 3.2 25.0% 4.0 8 20.0% 3.2 25.0% 4.0 Total 48.0 48.0 In the first scenario, the yield is 0.377, with a shutout in 66.5% of the simulations. In the second, the yield plunges to 0.154, and the shutout happens 85.5% of the time! This ought to be a concern. Particularly with the handful of candidates who were perched on the precipice of election now cleared, similar consensus might be harder to come by now, as it’s not simply “next man up” for who gets elected (acolytes of Allen and Hodges in particular can testify to that). It’s not hard to imagine a Classic Baseball slate containing such disparate candidates as long-dead Negro Leaguers whom nobody on the committee witnessed first-hand and whose statistics are incomplete (say, barnstorming pioneer John Donaldson or fireballer Dick Redding, whose career crossed from the pre-Negro Leagues Black baseball era into that of the major Negro Leagues) alongside still-living ones who the voters remember vividly and for whom the visual and statistical records are more fleshed out (Luis Tiant, perhaps). The process could easily grind to a halt without anybody honored. Some amount of polarization is necessary to elect at least one candidate. To show this another way, here’s a table of probability of a single candidate getting 12 votes out of 16 (75%) when he has an X% chance of being on any random ballot in an eight-vote format, independent of the odds of the other candidates. Probability of 12+ Votes Based on X% Ballot Odds pRandom Ballot At Least 12 Votes 99% 100.0% 96% 100.0% 93% 99.6% 90% 98.3% 87% 95.3% 84% 90.1% 81% 82.7% 78% 73.5% 75% 63.0% 72% 52.1% 69% 41.5% 66% 31.9% 63% 23.5% 60% 16.7% 57% 11.3% 54% 7.4% 51% 4.6% 48% 2.7% 45% 1.5% 42% 0.8% 39% 0.4% 36% 0.2% 33% 0.1% 30% 0.0% 27% 0.0% 24% 0.0% 21% 0.0% 18% 0.0% 15% 0.0% 12% 0.0% 9% 0.0% 6% 0.0% 3% 0.0% In graph form, that looks like a titration curve from a college chemistry lab: As you can see, at either extreme, a change in the odds has no effect, but in the middle, the odds of election quickly increase, nearly tripling as the individual ballot odds climb from 57% to 66% and then nearly doubling as they climb from 66% to 75%. Obviously, we can’t model every scenario. Still, I hope that this exercise helps to convey how the changes to the process, even if they’re well-intentioned — and I believe that the continued re-evaluation of the segregation-era candidates is laudable — actually make it much harder to produce honorees and increases the likelihood of the shutouts that frustrated observers and led to the Hall rejiggering the process in the first place. This is not to suggest that having a substandard honoree is better than having none at all, and that the process must be re-engineered to produce one every time (say, a runoff between the top candidates along the lines of what BBWAA voters did a few times). I feel confident that my two decades of evaluating committee processes amply illustrates the continued presence of strong but overlooked candidates who land on committee ballots, not that their mere presence guarantees optimal outcomes. We should still hope for processes that preserve the likelihood of such candidates being recognized, but with Dan’s help, I believe we’ve demonstrated that what’s about to be put in place decreases those odds.