Monday, December 28, 2015

BTRTN's Annual MLB Hall of Fame Projection: Look For Two New Entrants in January, 2016

The Hall of Fame voters came through with flying colors last year, electing four new members (Craig Biggio, Randy Johnson, Pedro Martinez and John Smoltz) and thus helping to break the logjam (a bit) caused by the unhappy confluence of the ten-player limit to voters and the Steroids Era.  There are five players on the ballot that have been tainted by the PED scandal:  Roger Clemens, Barry Bonds, Mark McGwire, Sammy Sosa and Gary Sheffield (or seven if you count Mike Piazza and Jeff Bagwell, for whom the whispers have been loud but the evidence scant).  Voters are split on the legitimacy of these HOF contenders; these five players receive plenty of votes (563 in total last year, or 12% of all votes) from those who choose to overlook their scandal.  They receive neither enough votes to get elected, nor so few as to be eliminated from future consideration (though Sosa may be this year as Rafael Palmiero was several years) -- but instead annually garner their 500+ votes, potentially robbing others from their place in Cooperstown.  So they are dragged along in their purgatory, embarrassments for years to come.

This year there are 15 newcomers to the HOF ballot who join 17 holdovers for a field of 32.  This analysis will focus on answering two questions:  1) who will be elected to the Hall of Fame this year, and 2) how many votes will everyone on the ballot receive?

To answer these questions, I scrapped my old method and have developed a brand new regression equation for newcomers, based on WAR (Wins Above Replacement value), and prior voting history for holdovers.  (PED tainting is also a variable in the equation.)  Had I used this new approach last year, my predictions would have been even more accurate than they were. 

One note:  the Baseball Gods are thwarting me and my new methodology a bit, because this year there will be 74 fewer HOF ballots as a result of a decision to reduce the number of members of the Baseball Writers Association of America who have voting privileges.  Because of this dramatic change in the composition of the voters, this could render the methodology I follow less effective.  While the general thinking is that the remaining writers will tend to be a little more “advanced stats” oriented and will also show a bit more leniency to the steroids gang, I have not attempted to incorporate such notions into the models.  (And one other note:  I have not looked at any early voting numbers at all.  I ran my models several months ago.)

So, armed with all this, let’s break down all 32 players on the ballot:

·         There will be two new members of the Hall of Fame: Ken Griffey, Jr. and Mike Piazza.  This is Junior’s first year on the ballot and he is a sure thing, racking up magnificent stats in an untainted Steroids Era career.  The HOF has never had a unanimous pick, and Griffey will not be one either, but I think he will get 97% of the vote.  Piazza has been the victim of a steroids whisper campaign but no evidence has ever surfaced to support that he was a PED user, and he is, hands down, the great offensive catcher of all time.  But it will be very close…the model projects Piazza getting exactly 75% of the vote, just what he needs. 

·         A total of 17 players will not gain election but will surpass the 5%   threshold for inclusion on next year’s ballot, although one of them, Alan Trammell, is now in his 15th and last year on the ballot and thus will not be back (a shame, he deserves to be in the HOF).  Jeff Bagwell and Tim Raines lead this list, and will fall just short of election.  Four players  in this group are tainted by the PED scandal, Roger Clemens, Barry Bonds, Mark McGwire and Gary Sheffield.  And three players are on the ballot for the first time, Trevor Hoffman, Billy Wagner and Jim Edmonds.

·         That leaves 13 players who will miss the 5% cutoff, 11 of them first-timers.  All of them were very good players, most with multiple All Star selections (and all with at least one), but none are close to being HOF material, with WAR’s generally below 50 and most in the 20’s.  Sammy Sosa, another PED-challenged player, and Nomar Garciaparra, are the two holdovers who will miss the cut.

The HOF seems to have done a better job of scrubbing the ballot of totally undeserving players.  Only one player on the ballot has a WAR of less than 20, Brad Ausmus.  Last year there were five such players.

My projection for voting percentage totals is shown in the chart below.  In the past I have relied on a regression equation that produced what I called the “TG Score,” which used On Base Plus Slugging Percentage Plus (OPS versus the league average over a career)) and career hits as the variable for hitters, and ERA Plus (ERA over the league average over a career) and innings pitched for pitchers.  Thus every player was measured on both “greatness” (OPS+ and ERA+) as well as “longevity” (hits and innings pitched).  The equation yielded a pretty good projection of percentage of votes attained, good enough last year to “beat” the legendary HOF predictor Bill Deane by a good margin.

But I noticed that my “TG Score” was highly correlated with WAR and, in fact, WAR is actually a better predictor of Hall of Fame votes than the TG Score.  So (of course) I created a new regression equation using WAR.  I use the WAR equation to predict the percentages for those players on the ballot for the first time.  For those who are holdovers, I use another prediction method based on their prior balloting history -- one method for second years, and another method for those who have been on the ballot for more than two years.

When I back-validated the new methodologies and re-predicted 2015, the results were superior to my “TG Score” model.  (I’ve included the comparisons at the end of this article for those of you that are interested.)

So, here are my predictions for the January, 2016 Hall of Fame announcement.  The columns are:  WAR = Wins Above Replacement value, YOB = Years on Ballot, PED = steroid issue, PY% is the percentage a playeri received in last year’s vote, and then my projection in the last column.  That is, to be absolutely clear, I am predicting that Ken Griffey, Jr. will receive 97% of the vote, Mike Piazza 75%, and so on.

2016
WAR
YOB
PED
PY%
Projected %
Ken Griffey, Jr.
83.6
1


97%
59.4
4

70%
75%
79.6
6

56%
60%
69.1
9

55%
59%
79.9
4

39%
42%
140.3
4
1
38%
38%
162.4
4
1
37%
37%
29.6
14

30%
32%
Trevor Hoffman
28.4
1

0%
30%
68.3
7

27%
29%
83.0
3

25%
26%
70.4
15

25%
26%
55.2
3

14%
14%
Fred McGriff
52.4
7

13%
13%
Larry Walker
72.6
6

12%
12%
Gary Sheffield
60.3
2
1
12%
11%
Mark McGwire
62.0
6
1
10%
8%
Billy Wagner
28.1
1

0%
8%
Jim Edmonds
60.3
1

0%
6%
Nomar Garciaparra
44.2
2

6%
4%
Sammy Sosa
58.4
4
1
7%
4%
Jason Kendall
41.5
1


3%
Troy Glaus
37.9
1


2%
Mike Hampton
29.0
1


2%
Garret Anderson
25.6
1


2%
Mike Lowell
24.8
1


2%
Mike Sweeney
24.7
1


2%
Luis Castillo
28.9
1


1%
Randy Winn
27.5
1


1%
Mark Grudzielanek
26.3
1


1%
David Eckstein
20.8
1


1%
Brad Ausmus
16.4
1


0%

The projected numbers speak for themselves, but here is some commentary on the candidates:

Our new Hall of Famers:

·         Ken Griffey, Jr: is a legend of our time, a man who compiled near-Bonds like statistics without the benefit of any enhancers, and with injuries that turned him from a superstar into a merely competent player for the second half of his career.  He had 630 homers, 1836 RBI and an 84 WAR.  A no brainer.

·         Mike Piazza also should have been a first-ballot HOF’er, but he will squeak in on his fourth try.  He is the greatest hitting catcher of all time, with 427 homers and 1,335 RBI.  This one is not even close, but for the whispering campaign about his possible usage of PED’s.

My own opinion is that the following players would be on my ballot, if I was a voter, that is, they should be in the Hall of Fame:

·         Jeff Bagwell’s stats are well above those of the average first baseman in the Hall of Fame, with a WAR of 80, but, like Piazza, he has suffered from PED rumors.  He will not get into the HOF this year, but he could next year with a little less competition (the newcomers in 2017 will include Ivan Rodriguez and Manny Ramirez, who have PED issues, and Vladimir Guerrero….in short, no “no-brainers.”)

·         Tim Raines is undervalued as a speedster; his 808 steals do not get incorporated well into the advanced stats.  The cocaine rap doesn’t help him either.  But Rock has a WAR of 69 and easily deserves to be in the HOF and will someday, perhaps even next year.

·         Curt Schilling’s ERA+ is a sterling 128 and his WAR is a hefty 81.  And if you like postseason performances, his stats are phenomenal:  11-2 with a 2.23 ERA.

·         Edgar Martinez gets dissed as a virtually full-time DH, but with an extremely healthy WAR of 68 he serves to be in the HOF; unfortunately, he won’t get in anytime soon.

·         Mike Mussina, in my view, was a stronger candidate than Tom Glavine; he has a better ERA+, won-loss percentage and WAR than Glavine, but Glavine demonstrated the mystical power of the 300-win mark.  It was ludicrous that Glavine received 92% of the votes to Mussina’s 20% in January, 2014.  Moose should make it, but he will be denied for some time to come.

·         Alan Trammell is on the ballot for the last time.  I’ve long thought he was worthy.  His stats are practically identical with the average of all HOF shortstops, except for those 185 homers which are nearly double his HOF shortstop peers.  Trammell was outshone in his career by Cal Ripken, Jr., and then eclipsed by the Jeter/A.Rod/Nomar trio of outstanding hitting shortstops.  But he belongs, for sure (as does his keystone partner, Lou Whitaker, who did not even receive the 5% mark in his first year of eligibility).

·         Jeff Kent has a WAR of “only” 55, but he is hard to deny given that he is all-time leading home run hitter among second baseman, and is third in RBI behind Rogers Hornsby and Napoleon Lajoie.  He was simply one of the greatest power-hitting second basemen ever and the greatest of modern times.  I’ve often thought that Lou Whitaker and Bobby Grich were HOF-worthy and Kent’s stats exceeds theirs.  He should be in, but won’t make it this year.

In my view, these players are not Hall of Famers, though they are worthy of strong consideration:

·         Fred McGriff has always been a very tough call for me.  His stats generally fall in line with the “borderline” HOF first basemen, with a WAR of 52.  As stated, Bagwell has an 80 WAR.  This is not a good era to be a borderline call.  Sorry, Fred.

·          Larry Walker suffers from “Coors Field Syndrome,” with an otherworldly home OPS of 1.068 and a merely very good .865 on the road.  That basically means that if he had played his career in any other ballpark, he would have been a player much like, say, Bobby Abreu or Ellis Burks, with a WAR in the 50’s instead of the 72 that Coors Field enabled him to achieve.  So I say no to Larry.  {Correction:  a reader pointed out that WAR is park-adjusted, which is correct, and thus Walker's 72 WAR already accounts for the Coors effect.  See the comments section below for the back and forth on this.  I'll rethink Walker next year!}

·         Jim Edmunds was a better player than most people think, with four All Star nods, two Top 10 MVP finishes, a slew of phenomenal catches and a surprisingly high WAR of 60.  That puts him on the borderline for HOF consideration.  I think he will just edge over the 5% mark and become a holdover.  But I don’t think he is HOF worthy and won’t last beyond next year if he makes it to the 2017 ballot.

·         Nomar Garciaparra, once upon a time, looked like a certain Hall of Famer.  He outshone Derek Jeter offensively in those early years.  But like Don Mattingly, injuries robbed him of the second half of his career and put him short of the mark.  He will likely fall off the ballot this year.

·         And that brings us to the most problematic candidates to predict, the first-ballot relievers Trevor Hoffman and Billy Wagner.  Relievers do not have a long track record of voting to learn from, and the nature of the closer role has evolved from the “two inning” guys (Gossage, Fingers and Sutter) to the “one inning” closer exemplified by Mariano Rivera.  In truth, relievers do not really have enough opportunities to have a high WAR; in truth, the role itself is overrated.  Mariano Rivera really stands alone in the role.  He has a WAR of 57; the next highest relievers are Rich Gossage and Hoyt Wilhelm at 40.

The rule of thumb for Hall of Fame status is a WAR of 60.   In modern times, players with WAR’s of 60-69 average a solid 40% first ballot voting percentage.  In contrast, those with WAR’s of 50-59 average only 6%.  Thus the dilemma for Hoffman and Wagner, each of whom have WAR’s of 28.  The model would indicate a WAR of 28 should translate to less than 5% of the vote.  But history tells us that relievers do better than they “should” in HOF voting.  Lee Smith has a WAR of 30, and consistently gets about 30% of the vote.

So, I’m eyeballing (that is, not modeling) a prediction of 30% for Hoffman (he has over 600 saves, the only player within hailing distance of Mariano), and 10% for Wagner, because the savvier voters (a subset of the whole) will recognize that he was every bit as good as Hoffman and Smith.  But I don’t think either should be in the Hall of Fame.  Only Mariano Rivera is truly deserving among the reliever set.  Apart from the 57 WAR, Mariano was superlative in his extensive postseason resume, whereas the others were all shaky.

·         Therefore I don’t think Lee Smith belongs in the HOF either, with his low WAR (30), even with his 478 saves.  And he is not budging much from his roughly 30% of the ballot range in this, his 14th year on the ballotSmith, Hoffman and Wagner are all in the same ballpark, and should be on the outside looking in.

As for the PED gang…

·         Nobody doubts the statistical credentials of Roger Clemens, Barry Bonds and Gary Sheffield.  Some believe Mark McGwire and Sammy Sosa do not have Hall-worthy stats, McGwire has 62 WAR and Sosa has a 58 WAR, which are borderline.  But the steroids rap voids them all from consideration, in my view.

And for the rest…

·         All the others are fairly forgettable by Hall of Fame standards, with WARS between 16 and 41.  Mike Sweeney made the All Star team five times; Troy Glaus and Mike Lowell made four; Garret Anderson made three and had a Top 10 MVP finish.  But these accomplishments, and the less ones of the others, are far short of HOF consideration.

So, look for smiles from Ken Griffey, Jr. and Mike Piazza on January 6th, and sighs from Jeff Bagwell and Tim Raines and all the rest. 

**************************************************************

Read on if you want to read about back validating!

The chart below attempt to do a few things.  First, as mentioned above, I managed to out-forecast the legendary Bill Deane last year, in that my projected votes were closer to the mark than his.  Deane is the former research associate at the Baseball Hall of Fame in Cooperstown and he has been doing HOF forecasting for 34 years, usually quite accurately.

You can tell that I did a bit better because his “total variance”, in the second from the right column, was greater than mine, the column to the left of his.  He was off by 112 (or about 3.3 percentage points per player), whereas I was off by “only” 80 (or 2.4 per player).  Bill had a great deal of trouble with first-balloters Pedro Martinez and John Smoltz.  I had some trouble with Smoltz, too, he was quite a unique candidate, with only Dennis Martinez as a relevant comparison. 

But, second, while I was pleased to beat the Master, I am more pleased by my brand new model.  The variance of this model, had I developed and used it last year, was only 58 (or 1.7 percentage points per player).  It did marginally better on Smoltz, but tightened up the variance on a number of others (notably Bagwell and Kent).

So, I hope to do better this year….noting again the fact that the voter base is a bit different last in prior years with the “downsizing” of the BBWAA voter rolls.


Jan. 2015

Projection (%)

              Absolute Difference (pp)
TG
Bill Deane
TG
TG
Bill Deane
TG
Original
Original
New Model
Actual
Original
Original
New Model
              96
              94
97
97
1
                3
0
              95
              57
95
91
4
              34
4
              65
              46
67
83
18
              37
16
              83
              79
82
83
0
                4
1
              68
              69
68
70
2
                1
2
              65
              60
59
56
9
                4
3
              50
              52
50
55
5
                3
5
              30
              36
32
39
9
                3
7
              33
              38
37
38
5
               -  
1
              33
              38
36
37
4
                1
1
              30
              32
33
30
0
                2
3
              27
              28
28
27
0
                1
1
              25
              27
20
25
0
                2
5
              24
              25
23
25
1
               -  
2
              23
              17
15
14
9
                3
1
Fred McGriff
              14
              14
13
13
1
                1
0
Gary Sheffield
              18
                5
16
12
6
                7
4
Larry Walker
              12
              12
11
12
0
               -  
1
Mark McGwire
                9
              13
10
10
1
                3
0
Don Mattingly
                7
              11
9
9
2
                2
0
Sammy Sosa
                4
                8
6
7
3
                1
1
80
112
58