World Series time! Enjoy Premium-level access to most features through the end of the Series!
October 16, 1999
The Top Pitchers of the 1990s
A Support-Neutral Approach
Who were the best pitchers of the 1990s? Almost everyone would rank Greg Maddux number one, and you probably wouldn't get too much argument by picking Roger Clemens number two, but after that it gets a little more controversial. With the voting for the Team of the 1990s in the home stretch, we thought it would be interesting to see who the Support-Neutral measures rate as the top starting pitchers of the 90's.
Support-Neutral Wins (SNW) and Support-Neutral Losses (SNL) are the expected number of wins and losses a pitcher's starts would produce given league average support from his offense and bullpen. We won't go into detail here about what the stats are and how they're calculated. If you're interested in finding out more, we have a short summary and a long article that describe the measures and compare them to other approaches to pitching evaluation.
We'll rank the pitchers based on the number of Support-Neutral Wins they had above what a replacement level starter would have had in the same number of Support-Neutral decisions. This measure is called Support-Neutral Wins Above Replacement, or SNWAR. We'll use a .425 winning percentage pitcher for our replacement level.
[Note: We don't actually have the Support-Neutral numbers for 1990 and 1991, because we don't have the game-by-game pitching lines necessary to calculate them. Instead we'll use an estimate based on innings pitched and earned runs allowed.]
With that background, here are the 10 best starting pitchers of the 1990s:
SNW SNL SNPct SNWAR 1. Greg Maddux 172.8 79.6 .685 65.5 2. Roger Clemens 158.9 78.1 .670 58.1 3. Randy Johnson 141.3 79.3 .641 47.6 4. Tom Glavine 148.2 93.0 .615 45.7 5. Kevin Appier 130.2 73.7 .639 43.6 6. David Cone 133.6 81.9 .620 42.0 7. Kevin Brown 140.9 96.6 .593 40.0 8. Mike Mussina 121.6 70.5 .633 40.0 9. John Smoltz 137.0 94.8 .591 38.5 10. Chuck Finley 134.7 102.2 .569 34.0
There aren't many names or placements on this list that will surprise people. All of these pitchers have been active throughout all or most of the 1990s, and all have been among the league's best for most of that time. If there are any surprises here, they're that Kevin Appier finishes as high as fifth and that Chuck Finley is one of the top 10.
Appier is underappreciated for a number of reasons. Until July, he had toiled in obscurity for a crummy Royals team. That crummy team hasn't given him any run support, so his actual record (120-90) understates the value of his pitching (130-74 Support-Neutral record) by a fair amount. And of the pitchers in the top 10, he's been the least effective recently, thanks to injury.
But none of that should lessen our appreciation of Appier's excellence in the middle years of the decade. According to his Support-Neutral record, he was the third-best starter in the AL in 1992, the best by a wide margin in 1993, fifth in 1994, fourth in 1995, eighth in 1996 and eighth in 1997.
Perhaps more interesting than the top 10 is a list of the "also-rans", numbers 11 through 20:
SNW SNL SNPct SNWAR 11. Pedro Martinez 90.7 45.1 .668 33.0 12. Curt Schilling 105.8 69.6 .603 31.2 13. Andy Benes 123.7 105.5 .540 26.3 14. David Wells 113.5 92.1 .552 26.1 15. Jimmy Key 94.3 66.3 .587 26.1 16. Dennis Martinez 93.0 66.3 .584 25.3 17. Jose Rijo 75.2 43.3 .634 24.8 18. Jack McDowell 98.8 75.3 .567 24.8 19. Alex Fernandez 104.1 82.7 .557 24.7 20. Bret Saberhagen 82.0 53.7 .604 24.3
This list contains more surprising names because many of them are no longer pitching, or no longer pitching effectively. The great performances by Jimmy Key and Dennis Martinez in the early years of the decade aren't as fresh in our memories. Jose Rijo was practically as good in the first half of the decade as Pedro Martinez has been in the second half.
Speaking of Martinez, why is he here instead of in the top 10? Only because of his late start; he didn't become a rotation starter until 1994. Pedro hasn't pitched enough to accumulate the wins above replacement value that some of the full-decade pitchers did. But if you look at his Support Neutral Winning Percentage (SNPct), which measures the rate at which a pitcher contributes expected wins to his team, Martinez is among the elite of the 1990s, just below Maddux and Clemens.
While we were at this, we also generated a quick list of the top relievers of the 1990s. This ranking is not based on Support-Neutral records, which only apply to starting pitchers. Instead it is based on a modified version of Pete Palmer's Adjusted Pitching Wins (APW), which uses run prevention to estimate wins added compared to an average pitcher. Unlike the starter rankings, this list only evaluates 1990 through 1998.
IP ER APW 1. John Wetteland 536.3 151 9.8 2. Jeff Montgomery 643.3 225 9.0 3. Roberto Hernandez 508.7 170 7.6 4. Jeff Brantley 656.3 214 7.0 5. Mike Jackson 627.7 218 6.8 6. Mark Eichhorn 428.0 133 6.7 7. Rick Aguilera 627.3 249 6.7 8. Tom Henke 347.3 96 6.5 9. Eric Plunk 678.0 261 6.5 10. Trevor Hoffman 441.7 136 6.4
Once 1999 gets added in, there will be a few changes to the rankings. Roberto Hernandez might pass Jeff Montgomery for second, Mike Jackson would probably pass Jeff Brantley, Trevor Hoffman would jump up a few spots and Mariano Rivera would probably crack the top 10.
We were surprised by several of the names on this list. If you ask a random baseball fan to name the top relievers of the decade, spelling out the rough criteria that were used to make this list--lots of innings, few runs allowed, sustained performance across the decade rather than good individual seasons--you're not going to hear the names Jeff Brantley, Mark Eichhorn or Eric Plunk very often. The more memorable relievers are those who have had dominating stretches--Rob Dibble, Randy Myers, Dennis Eckersley, Bryan Harvey, Mark Wohlers--but so many of them seem to fall off the face of the earth after two or three years. It's rare to find relievers like Wetteland, Hernandez and Jackson who sustain their high levels of performance over an extended period.