Predicted Outcome of the 2015 BBWAA Hall-of-Fame Voting
This is my third year in which I have written about the BBWAA Hall-of-Fame ballot, including articles about every player on the ballot.
This year, I thought I'd try something new as well: predicting Hall-of-Fame voting.
I'll spare the suspense and start off with my final answer, before some analysis of my results and explaining how I got there. The first table below presents four sets of numbers for the players on the 2015 BBWAA Hall-of-Fame ballot: predicted 2015 vote percentages, actual 2014 vote percentages, estimated 2014 vote percentages in the absence of the ballot cap, and actual 2013 vote percentages.
Summary of Results
If my forecast turns out to be completely correct (which is, of course, highly, highly unlikely), three players will be elected to the Hall of Fame from this ballot, with two other players coming relatively close. Re-ordering the above table, and limiting it to players who I predict to receive at least 5% of the vote (and thereby remain on the 2016 Hall-of-Fame ballot), I get the following.
I think of this as dividing into four groups of players - which are color-coded above.
The top group are those who I predict will be elected to the Hall of Fame this year: Randy Johnson, Pedro Martinez, and Craig Biggio.
The second group of players are those who appear to be on or near the cusp of induction: John Smoltz, Mike Piazza, Jeff Bagwell, and Tim Raines.
The first two players in this group - Smoltz and Piazza - are close enough to 75% that I think there is at least some chance that they are elected this year. Based on public results so far, I could well be under-estimating John Smoltz's likely vote total, and I would not be terribly surprised if he was elected this year. I would be somewhat more surprised - albeit pleasantly so - to see Piazza make it this year, although I do expect him to be elected within the next two to three years.
I would be shocked if either Bagwell or Raines were actually elected this year, but if my predictions of their vote totals are reasonably accurate, they could both be poised to move toward eventual induction over the next 3 - 4 years (although I believe Raines now only has two more years on the ballot after this one).
The third group of players are those who are hanging around, below 50%, but with significant enough support that they are, at a minimum, likely to stay on the ballot for the foreseeable future, and may even be able to begin to move toward election.
The final group of players seem unlikely, to me, to be likely to gain any votes in the future much less move toward election by the BBWAA.
Taking into account time on the ballot, there is, perhaps, some crossover between these last two groups. Specifically, it seems clear to me that Lee Smith and Alan Trammell have peaked in BBWAA support and clearly belong in this final group, merely biding time until they fall off of the BBWAA ballot, while Jeff Kent has enough time left on the BBWAA ballot that, although his predicted vote total here is lower than both Trammell's and Smith's, Kent's vote total could increase in the future (although I would be surprised if he were actually elected by the BBWAA).
Having presented my forecast, the next section of the article describes its derivation.
Derivation of My Forecast
The source data which I used to derive my Hall-of-Fame vote forecast were historical Hall-of-Fame voting totals - most prominently actual vote totals for the returning candidates on this ballot from 2013 and 2014. In addition, I relied upon my own earlier analysis of what 2014 voting results would have looked like in the absence of the ballot cap. I am also aware of the Hall-of-Fame Ballot Collecting Gizmo at Baseball Think Factory which reports partial results for the 2015 electoin for voters who have been willing to share their votes. I tried to limit my reliance on this, as the pool of publicly available ballots is not necessarily a random statistical sample of all voters but it did influence my results in a couple of specific instances.
The starting point for my analysis was 2014 vote totals. The target of my forecast was vote totals (e.g., 456 votes for Craig Biggio), not vote percentages (e.g., 84.0% for Craig Biggio).
Voting Pool
In 2014, 571 voters cast a total of 4,793 votes (8.39 votes per ballot). Of these 571 voters, 299 votes have been made public on Ryan Thibs' wonderful Hall-of-Fame tracking spreadsheet.
For the 2015 election, the Hall of Fame and the BBWAA changed the voting proceduress somewhat. While they failed to make the obvious change that they should have - eliminating or at least raising the 10-man ballot limit - they did make a couple of changes that may have an effect on the size of the electorate. Specifically, "voters will now be required to complete a registration form and sign a code of conduct." In addition, "names of those BBWAA members casting Hall of Fame ballots will now be made public with the election results" (although "an individual's ballot will not be revealed by the Hall of Fame").
It is hard to know precisely what, if any, impact this might have on the size and composition of the electorate in the 2015 Hall-of-Fame election. But that isn't going to stop me from making a (hopefully educated) guess. One would hope that the requirement to sign the code of conduct would not have any effect on the electorate - although this rule was clearly passed in response to a specific 2014 voter, Dan LeBetard. The other two changes - registration and public release of the names of all voters - could discourage some voters from participating, however. One would guess that voters who have made their ballots public in the past would be unlikely to be dissuaded from voting by these new rules, but some voters who have been reluctant to publicly reveal their vote in the past may be similarly reluctant to be publicly identified as voters in the future.
As I noted above, of the 571 voters in the 2014 Hall-of-Fame election, 299 eventually made their results public to Ryan Thibs. This means, of course, that 272 ballots remain anonymous. For my forecast here, I made two assumptions.
1. 10% of anonymous 2014 voters will not vote in 2015
2. These voters' ballots were equivalent to the average non-public 2014 ballot
As a technical detail, I did my projections based on a somewhat earlier version of Thibs' 2014 spreadsheet that had 282 anonymous votes - 10% of which are 28 voters, so I ended up with a projected vote total of 543.
I am aware of three voters who did reveal their ballot publicly last year, but who will not be voting this year. One is Dan LeBetard, whose voting privileges were stripped by the BBWAA for his having given away his vote. The other two - Buster Olney and Lynn Henning - publicly announced their intention to abstain in protest of the current voting procedure (specifically, both object to the 10-man ballot cap - as all good and right-thinking people should). I have not taken these three votes into account in my analysis, nor have I taken into account the new voters this year who are eligible for Hall-of-Fame voting for the first time. Nor, for that matter, have I attempted to account for any old voters who may have died in 2014. My forecast here is not being built up by guessing the votes of all possible voters and adding them up. I am, instead, assuming that the sorts of changes in the mix of voters described in this paragraph will more or less cancel each other out.
Returning Players
The first step in my analysis involved looking at returning players. For the 17 players returning to the ballot from last year, I compared their vote totals in 2013 and 2014 - using my estimated uncapped percentages - and attempted to identify trends in their vote totals.
Based on this analysis, I placed players into one of four categories: "steroid users", players trending upward, players trending downward, and players not trending.
Two returning players, Jeff Kent and Mike Mussina, were new to the ballot last year, so had no trend from 2013. I placed both Kent and Mussina in the last category, basically by default. Joining Kent and Mussina in the "no trend" group were Jeff Bagwell, Edgar Martinez, and Curt Schilling.
In the last of these cases, Schilling's vote percentage actually declined from 2013 (38.8%) to 2014 (estimated 35.6%, uncapped). My hypothesis is that this decline was due to the debut of three strong pitching candidates in 2014 - Tom Glavine, Greg Maddux, and Mike Mussina. While three strong pitching candidates also debut in 2015 - Randy Johnson, Pedro Martinez, and John Smoltz - this is not a change from 2014, so I would not expect a further decline in Schilling's vote total. On the other hand, Schilling remains no better than the third-best pitching candidate on the current ballot, with Johnson and Martinez replacing Maddux and Glavine as clearly better, "steroid-free" candidates, so neither would I expect Schilling's (or Mussina's) vote percentage to increase much either.
The other three candidate groupings and my estimate of how their vote totals are likely to change are discussed next.
"Steroid Users"
The first group of players which I lumped together are those candidates whose vote totals are obviously being (significantly) negatively affected by their alleged use of steroids or other performance enhancers. There are four such players returning to the 2015 Hall-of-Fame ballot: Barry Bonds, Roger Clemens, Mark McGwire, and Sammy Sosa.
The online baseball circles which I frequent (Baseball Think Factory and Twitter) tend to look less unfavorably on steroid use and tend not to see it as a disqualifier for Hall-of-Fame membership. There also tends to be a prevailing opinion that society's views in general and baseball writers' views in particular of steroid use - especially pre-testing (2003-04) - are going to ease over time, so that, eventually, it is all but certain that "steroid users" will find their way into the Hall of Fame.
I tend to disagree. I have seen no evidence that the BBWAA electorate's view of steroids is softening over time. Mark McGwire's Hall-of-Fame vote totals are probably the best measure of this. Since his debut on the Hall-of-Fame ballot in 2007, his vote totals have been (in order) 128-128-118-128-115-112-96-63. Even adjusting for the ballot cap, I estimate that his 2014 total would have been only 93. Rather than softening over time, the BBWAA's anti-steroid sentiment seems to have gotten stronger over the past two to three years after holding remarkably constant for many years. I also think the view that younger people are less intolerant of steroids is not true so much as the age cohort who were relatively young baseball fans during the "Steroid Era" are more tolerant of steroids. Last year, I asked my then-12-year-old son if he thought Barry Bonds should be in the Hall of Fame (note: I do believe that Bonds should be in the Hall of Fame). His response was probably two solid minutes of hysterical laughter. His generation of baseball fans have come of age at a time when steroid use is unambiguously illegal, tested for, and punished, and is universally condemned by the baseball media. It would not surprise me at all if Barry Bonds remains outside of the Hall of Fame for the next 50 years or more.
But that's opinion. What about fact? As I noted above, even controlling for the ballot cap, I estimate that Mark McGwire would have lost 3 Hall-of-Fame votes between 2013 (96, 16.9%) and 2014 (93, 16.3%). I estimate similar declines in support for Barry Bonds (36.2% to 35.9%) and Roger Clemens (37.6% to 36.6%), and a much more significant decline in support for Sammy Sosa (12.5% to 8.4%). Overall, the average change in support for these four players from 2013 to 2014 (estimated, uncapped) was 9.9 percent (a decline from 12.5% to 8.4% is a 32.6 percent decline - 8.4/ 12.5 - 1 = -32.6%). I assumed a similar decline for all four of these players in 2015.
This ends up assuming a larger decline for Bonds, Clemens, and McGwire in 2015 than in 2014, and a somewhat smaller decline for Sosa, which may not be reasonable. On the other hand, Comment #355 here indicates a 10% decline in announced support for both Bonds (48.1% to 44.4%) and Clemens (46.8% to 43.2%) in early sampling of public 2015 votes and much more dramatic declines for both McGwire (11.7% to 6.2%) and Sosa (7.8% to 4.9%).
Trending Upward
Comparing 2013 results to estimated 2014 results in the absence of a ballot cap, three players' vote percentage increased from 2013 to 2014: Craig Biggio, whose actual vote percentage increased from 68.2% to 74.8% (78.3% uncapped); Mike Piazza, whose actual vote percentage increased from 57.8% to 62.2% (64.4% uncapped); and Tim Raines, whose actual vote percentage declined from 52.2% to 46.1%, but whose vote total would have increased (by my estimate) to 53.1% had there been no ballot cap.
I debated treating Raines as a player "not trending", but decided to include him as "trending upward" because Ryan Thibs' HOF Tracker spreadsheet, where he is tracking voter-specific changes in player support, showed Raines having gained 7 votes through 92 public ballots (+7.6%).
The average gain from 2013 to (estimated uncapped) 2014 for Biggio, Piazza, and Raines was 9.3 percent. I assumed a similar gain for all three players in 2015.
Trending Downward
Finally, comparing 2013 results to (estimated uncapped) 2014 results, five players who are not starting pitchers and have not been associated with steroids saw non-trivial declines in support: Don Mattingly (from 13.2% to 10.0% uncapped, 8.2% actual), Fred McGriff (from 20.7% to 18.0% uncapped, 11.7% actual), Lee Smith (from 47.8% to 38.0% uncapped, 29.9% actual), Alan Trammell (from 33.6% to 27.8% uncapped, 20.8% actual), and Larry Walker (from 21.6% to 18.2% uncapped, 10.2% actual).
The average decline from 2013 to (estimated uncapped) 2014 for these five players was 18.1 percent. I assumed a similar decline for all five players in 2015.
New Candidates
So, to some extent, that was the easy half of the ballot. For new candidates, I tried to relate players to the players on the 2014 ballot most similar to them. In some cases, I also went back further in time to find good parallels. And I also relied on the results publicly released so far.
New players, in projected order of finish in voting:
Randy Johnson - My first guess for Randy Johnson is a simple average of the 2014 vote totals of Greg Maddux and Tom Glavine, 94.6%.
Pedro Martinez - My first thought was to use Sandy Koufax's total the year he was elected (1972), 86.9%. Based on public results, this seemed low, so I decided to take the average of my guess for Randy Johnson (94.6%) and Sandy Koufax (86.9%), 90.7%.
John Smoltz - To be honest, I'm not quite sure what to make of John Smoltz.
In my mind, there is a clear hierarchy among pitchers of the 1990's and 2000's. Based purely on statistics, the top two pitchers of the era were clearly Roger Clemens and Greg Maddux, followed by the two other pitchers debuting on this year's ballot, Johnson and Martinez. In my mind, Tom Glavine is clearly next in line (in terms of perception), thanks to the 300 wins and two Cy Young awards.
Below those five, then, the other three starting pitcher candidates on this ballot - Mussina, Schilling, and Smoltz - have always seemed to me like three peas in a pod: they could plausibly be put in any order and one could reasonably draw one's Hall-of-Fame in/out line either above or below them, but it seems very hard to me to figure out how to draw that line between them.
But starting last year, I noticed a very clear difference in the discussion around Schilling and Mussina - as plausible Hall-of-Fame candidates - and Smoltz, who was almost universally referred to by baseball media folks as a future Hall-of-Famer.
Based on that, my first guess was to take the simple average of Schilling's (29.2%) and Glavine's (91.9%) 2014 vote totals as a first guess for Smoltz - which is a pretty wide range, but one that reflects the wide level of uncertainty I have about Smoltz's Hall-of-Fame candidacy; this works out to 60.6%. Through 104 ballots, though, the Ballot Collecting Gizmo shows Smoltz at 89.4% - and holding steady that high.
I'm still not entirely sure I get that - that's higher than Sandy Koufax got and barely less than Smoltz's teammate Tom Glavine got. But I decided to hedge my bets a little bit and bump up my forecast for him to a weighted average that was one part Schilling and two parts Glavine: 71.0%. That's still quite a bit below 89%, of course, and if I absolutely had to pick one of my predictions and guess the direction of my error, it would be that I'm under-estimating Smoltz's final vote total. But for now, I'm going with 71.0% for my first-round guess.
Gary Sheffield - The only other new candidate with a realistic chance of staying on the ballot beyond this year is Gary Sheffield. The player already on the ballot whose case seems the most similar to Sheffield's to me is Sammy Sosa - more than 500 HR's, but unpopular with the media, with vague steroid associations, and with no obvious advocates within the electorate. So, my first guess for Sheffield was that he would match Sammy Sosa's vote total, which, after adjusting Sosa down as described above, is 6.4%.
Carlos Delgado - My first thought on Carlos Delgado (473 HR's, 4 top-10 MVP finishes) was that his support could be similar to Fred McGriff (493 HR's, 6 top-10 MVP finishes) (9.6% after adjustments) or maybe half of McGriff's support (4.8%, which would put him on the cusp of hanging on for a second year). But through 104 ballots, the HOF Gizmo had only found 1 vote for Delgado (1.0%).
So, I decided that the best comp(s) for Delgado were probably the players on the 2014 ballot who didn't make the 5% cut. The top two vote-getters among players who debuted on the 2014 ballot and didn't make the cut were Moises Alou (6 votes) and Luis Gonzalez (5 votes). Delgado struck me as a stronger candidate than either Alou or Gonzalez, so I made my guess for Delgado the sum of Alou's and Gonzalez's support, 11 votes (2.0%).
Nomar Garciaparra - Garciaparra seems to me to have a comparable Hall-of-Fame case to Carlos Delgado - different (peak vs. career) - but comparable. So, I used the same guess for Garciaparra as Delgado, 2.0%.
Brian Giles - Brian Giles is probably the last candidate here for whom one could construct an even remotely plausible Hall-of-Fame candidacy. My first guess for Giles was Moises Alou's 2014 vote total, 6 (1.1%).
The 2014 Hall-of-Fame ballot was probably the most crowded with strong candidates of my lifetime. And yet, with a ballot that strong and a 10-man ballot cap, Eric Gagne and J.T. Snow were able to score 2 votes and Armando Benitez and Jacque Jones received one vote apiece. If J.T. Snow could get two votes, I don't see why Tony Clark couldn't do the same. Likewise, the Hall-of-Fame cases of Tom Gordon and Troy Percival seem no worse than Eric Gagne's case. So, basically, my estimates for the rest of the debut class of 2015 were the totals of Gagne, Snow, Benitez, and Jones, if any of these players seemed comparable.
Opening Ballot Spaces
Putting all of the above together, I came up with a first pass total of 4,128 votes by 543 voters, for a total of 7.60 votes per ballot. In the 2014 election, as shown in an earlier table, the total vote was 8.39 votes per ballot.
In 2014, in the absence of the ballot cap, however, I estimate that the total votes per ballot would have been 9.49.
Based on the lower average ballot size in 2015, it seemed reasonable to suspect that some voters who had full ballots in 2014 might have space to add additional votes in 2015 - e.g., somebody who voted for Maddux, Glavine, Thomas, and Jack Morris in 2014 (all of whom were named on over 60% of ballots and are no longer on the 2015 ballot) but only voted for Johnson, Martinez, and Smoltz among newcomers.
Adding 50% of my estimate of "capped" votes in 2014 (see here) produced 8.17 votes per ballot for the 2015 ballot. This number is still slightly lower than the 2014 total, but not terribly so. Adding in these previously capped votes produced the final estimates which I showed earlier.
If I Had a Vote
So, that's my prediction of how the actual BBWAA voters are likely to vote. And just in case anybody cares, here's how I would evaluate the players on the 2015 Hall-of-Fame ballot and who I would vote for. I'll start with a variation on the table I put together in my earlier article on the 2015 Hall-of-Fame ballot.
I've added an additional statistic here, which is the sum of my four basic value measures: pWins and eWins over both positional average and replacement level. The players are sorted by career value in this combined statistic, for which I have also included postseason statistics.
If I were being asked to rank all 34 of the above players, I might tweak the order of a handful of players in the above table, but that's basically a reasonable enough ranking of the players on the ballot by value (as measured by Player won-lost records).
If the ballot was completely open and just asked for a simple yes/no vote on every player on the ballot with no further restrictions, I would vote for everybody in the above table through Sammy Sosa and I might also consider voting for Fred McGriff, Carlos Delgado, and Nomar Garciaparra as well.
Unfortunately, Hall-of-Fame voting is not that simple. Instead, for reasons that completely baffle me, the Hall-of-Fame and/or the BBWAA have continued to limit their voters to choosing no more than 10 candidates.
Hence, Hall-of-Fame voting is not a simple yes/no worthy/unworthy voting exercise but is, instead, an exercise in game theory. This is true even if one chooses to "not vote strategically" which is, itself, a strategy - a valid and easily defensible strategy, certainly, but a "strategy" nonetheless.
For me, if I were an actual BBWAA voter, my goal in voting would be to get as many qualified candidates elected to the Hall of Fame as possible. Thinking back to the four tiers of candidates that I identified earlier, this means, as a first step, voting for everybody in tiers one and two. Alphabetically, that would be Bagwell, Biggio, Johnson, P. Martinez, Piazza, Raines, and Smoltz. That fills up 7 of the 10 allowable slots.
Moving down to tier three, I see two possible pairs of candidates: Bonds and Clemens or Mussina and Schilling. As I discussed above, based on voting history, I believe that it is extremely unlikely that Bonds and Clemens (and other "steroid" candidates) will ever be elected to the Hall of Fame through the BBWAA. In contrast, I believe that, with a reasonably strong showing this year, Schilling and Mussina could be well positioned to move toward eventual election by the BBWAA - especially if Smoltz is elected this year, leaving Schilling and Mussina as the only "steroid-free" starting pitcher candidates on next year's ballot (and for the foreseeable future). Hence, my "strategic" vote would be for Mussina and Schilling.
That fills up 9 of the 10 allowable slots.
For my last slot, I would choose to vote for my favorite player (who belongs in the Hall of Fame) among the remaining candidates. For me, that is Alan Trammell.
So, my final ballot, alphabetically, would be the following ten candidates.
All articles are written so that they pull data directly from the most recent version of the Player won-lost database. Hence, any numbers cited within these articles should automatically incorporate the most recent update to Player won-lost records. In some cases, however, the accompanying text may have been written based on previous versions of Player won-lost records. I apologize if this results in non-sensical text in any cases.
Home
List of Articles