Introducing JMWAR: using chatgpt to evaluate Baseball Hall of Fame voting
On January 23, 2024 a very important election took place, with the BBWAA announcing their inductees for the Hall of Fame, Adrian Beltre, Joe Mauer, and Todd Helton.
Voting for the Baseball Hall of Fame has improved over the past several years, but I think there are still ways to make it better.
The main reason for needing improvements is the popularization of the statistic WAR (wins above replacement), but by combing WAR with traditional stats and player accomplishments I have created a new stat, JMWAR, to determine which players are deserving of more consideration.
Therefore, JMWAR consists of the following components: JM(the author of the stat)W(ins above replacement)A(ccolades)R(eputation). After combining the amount of WAR and accolades acquired, a third component is used to adjust for the players involvement in various scandals.
Like the Hall of Fame Monitor, the point of JMWAR is to create a singular score where 100 or higher indicates a Hall of Famer. Some fans will argue that “either a player is Hall of Famer, or they’re not”, but it’s often not that simple with lots of gray area in between.
As I see it, there are four different categories for Hall of Fame candidates:
Slam Dunk, first ballot-hall of famer
Eventual Hall of Famer
Possible Hall of Famer, deserving of a few votes
Good enough to appear on the ballot
The line between each of these categories is somewhat blurred, which is why I created the point’s system (JMWAR) to determine which category a player falls into.
So for JMWAR the points system works like this:
10 points for each MVP, Rookie of the Year, or Triple Crown (batters or pitchers)
10 points for each World Series Win
5 points for each all-star appearance
5 points for each silver slugger award
5 points for each batting title
5 points each gold glove won at P, C, 2B, 3B, SS
3 points for each gold glove won at 1B or OF
10 points for 2000 hits, 20 points for 2500 hits and 100 points for 3000 hits
10 points for 300 HR, 20 points for 400 HR, and 30 points for 450 HR, 40 points for 500 HR, 50 points for 550 HR, and 100 points for 600 HR
10 points for a career AVG of .300, a career OBP of .400, or a career SLG of .500
20 points if the Career AVG is .320 or higher, 30 if it’s .340 or higher, 40 if it’s .360 or higher (Ty Cobb)
10 points for 1500 career RBI
10 points for 300 SB, 20 points for 600 SB, 100 points for 800 SB
10 points for hitting 60 home runs in a season or four in one game
10 points for 50 Career WAR, 20 points for 60 Career WAR, 30 points for 70 Career WAR, and 100 points for 100 Career WAR
Now for the pitching side of things:
10 points for each Cy Young award
5 points for 100 Wins, 10 points for 150 Wins, 20 points for 200 Wins, 30 points for 250 Wins, and 100 points for 300 wins
10 points for 1000 stikeouts, 20 points for 2000 strikeouts, 100 points for 3000 strikeouts
10 points for a career ERA below 3.00 (minimum 1,000 IP)
5 points for 100 saves, 10 points for 200 saves, 20 points for 300 saves, 50 points for 400 saves, and 100 points for 600 saves
5 points for winning the Rolaids Relief Award (because the best relievers only pitch at most 1/2 as much as the best starters)
10 points for a perfect game
5 points for each no-hitter that isn’t a perfect game (E.G. Nolan Ryan has 35 points)
10 points for a 20 strikeout game
At the point, I enlisted the help of ChatGPT to put this information in formulaic form. After a few tweeks, here’s what it came up with:
A little messy, but you get the point…
Afterwards, I asked ChatGPT to calculate the scores for each player, where it made numerous mistakes. To be fair, it had to use Bing search for information, as Web crawlers are blocked on Baseball Reference. After double-checking its work, I added the final component, subtracting points for reputational failures. And that leaves us with the final list of players on the 2024 Hall of Fame ballot in order:
Alex Rodriguez (A-Rod): 415 (515 before deduction, -100 points for suspension)
Adrian Beltre: 235
Manny Ramirez: 160 (260 before deduction, -100 points for multiple suspensions)
Todd Helton: 139
Gary Sheffield: 130 (-50 deduction for connection to BALCO)
Joe Mauer: 125
Francisco Rodriguez (K-Rod): 120
Carlos Beltran: 114 (164 before deduction, -50 points for Astrogate)
Mark Buehrle: 110
Billy Wagner: 100
Matt Holliday: 100
Andruw Jones: 100
Omar Vizquel: 100
Torii Hunter: 92
Andy Pettitte: 85 (135 before deduction, -50 points for steroids)
Jimmy Rollins: 80
Chase Utley: 80
Adrian Gonzalez: 67
David Wright: 65
Bobby Abreu: 58
Jose Bautista: 55
Victor Martinez: 45
Bartolo Colon: 20 (70 before deduction, -50 points for steroids)
It’s interesting that even with a huge point deduction for steroids, Alex Rodriguez and Manny Ramirez remain two of the best candidates. Either their point deduction should be larger, or they should be in Cooperstown. Futhermore, Matt Holliday should have received a lot more than just the meager 4 votes he got on the ballot. However, three of the highest players (Nos. 2,4, and 6) ended up getting elected, making JMWAR a good indicator of electibility. You can view the whole results here.