Roger Clemens ... does he have a case for non-support?

Tom Tippett
April 7, 1998

Two years ago, a caller to the Boston sports radio station decried the lack of support Roger Clemens had received from his bullpens over the years. It was early May, 1996, and the pen had already blown four leads in games Roger had started. His record was 1-4, and he deserved better. I wondered if the bullpen had been letting him down for years, or whether this was merely a brief run of bad luck.

So I wrote a little computer program to compile the results of all of Roger's career starts, and this is how his 359 career starts looked at the time:

• he was 75-21 in complete games
• in the 90 games in which he left with his team trailing, the Sox were 8-82, and Roger's record was 0-78. The team took him off the hook twelve times: eight wins and four games in which the team rallied to tie before a reliever took the loss.
• in 26 games when the score was tied when he departed, the Sox were 11-15
• he had handed the bullpen a lead 147 times, with the team winning 126 of those games. In addition to the 21 losses, there were 16 other games in which the pen gave up the lead before the Sox rallied to win.

In other words, the team had taken Clemens off the hook 12 times and had blown 37 leads. All in all, it seemed quite appropriate for Roger and his fans to be frustrated.

But what's normal? We hear lots of talk about saves and blown saves and save percentages, always from the perspective of the closer. But today's starters often hand the ball to middle relievers or setup men, and there may no longer be a save opportunity by the time the closer is summoned. How often does a starter get the victory after he leaves the game with a lead? And how often does he take the loss when he hands over a deficit? Does the record confirm that Roger was unlucky? Were there others who were even less fortunate?

Five-year Averages

To answer these questions, I updated the program and ran it for all pitchers for the five seasons from 1993-97. Let's start by looking at the overall averages for this period:

Margin   GS    TmW   TmL  TmPct  PchrW  PchrL  Blown  Pct   OffHook  Pct
------  ----  ----  ----  -----  -----  -----  -----  ----  -------  ----
  -5+   1650    72  1578   .044          1526                 124    .075
  -4    1338    87  1251   .065          1180                 158    .118
  -3    1649   175  1474   .106          1368                 281    .170
  -2    1853   282  1571   .152          1405                 448    .242
  -1    2063   492  1571   .238          1293                 770    .373
   0    2209   986  1223   .446           215                   
   1    2295  1530   765   .667   1080     74   1215  .529
   2    1863  1490   373   .800   1216     15    647  .347
   3    1453  1267   186   .872   1108           345  .237
   4     981   920    61   .938    838           143  .146
   5+   2030  2000    30   .985   1937            93  .046

 Trail  8553  1108  7445   .130                              1781    .208
 Ahead  8622  7207  1415   .836                 2443  .283

  CG    1446  1114   332   .770

I'll take a moment to work through this table, since it establishes the framework for the rest of this article:

• the Margin column refers to the run difference when the starter left the game. If he left the game with his team down by five or more runs, his start goes into the first row. If he was trailing by four, it goes in the second row. And so on.
• the Trail and Ahead rows show totals for all situations where the starter left while trailing or leading by any number of runs.
• the last row is for complete games. In these outings, the starter always gets the win or loss.
• the second column shows the number of starts falling into each category, and the next three columns show the team's wins, losses, and winning percentage in these starts.
• the PchrW and PchrL column shows the starting pitcher's record in these games.
• if the pitcher leaves with the lead, he can still take the loss if he's responsible for baserunners that score the go-ahead runs. More often, however, he'll either get a win or nothing. The Blown column and the one that follows it show how often he gets nothing.
• a starter can never be given the win if he leaves with his team tied or behind, so in these games, what's interesting is how often he gets tagged with the loss and how often his team rallies to take him off the hook. Those figures are shown in the last two columns.

This table gives us some context in which to evaluate Clemens' experience. It's not a perfect match, of course, since the Clemens numbers go back to 1984 and the table is for the past five years only. And it's reasonable to assume that in this era of higher offense, leads would be a little harder to hold and deficits a little easier to make up. But it still provides some useful benchmarks.

Clemens' Career Record

First, however, let's update Clemens' career numbers to include the past two seasons:

Margin   GS   TmW  TmL  TmPct  PchrW  PchrL  Blown  Pct   OffHook  Pct
------  ----  ---  ---  -----  -----  -----  -----  ----  -------  ----
  -5+    18     0   18   .000           18                   0     .000
  -4     16     0   16   .000           16                   0     .000
  -3     23     5   18   .217           18                   5     .217
  -2     25     1   24   .040           22                   3     .120
  -1     24     4   20   .167           17                   7     .292
   0     31    14   17   .452            1                   
   1     40    24   16   .600    17      2     21   .550
   2     32    27    5   .844    24             8   .250
   3     34    31    3   .912    27             7   .206
   4     21    20    1   .952    17             4   .191
   5+    44    44    0  1.000    44             0   .000

 Trail  106    10   96   .094           94                  15     .142
 Ahead  171   146   25   .854   129            42   .246

  CG    108    84   24   .778

Here are a few things I notice when I compare this table to the overall averages:

• Clemens' record in complete games seems remarkable at first glance, but is really quite normal. What's unusual is the large number of complete games -- 108 in 416 starts, or 26%, while the MLB average was 11% from 1984-97.
• his teams rarely bailed him out when he left while behind (.142 vs an average of .208). In 416 starts over 14 seasons, his mates have never rallied from a deficit of four or more runs, and were also well below average with 1- and 2-run margins. If he'd been taken off the hook at the average rate (.208), he'd have about 7 fewer losses.
• his teams have done reasonably well when he left them with a lead. His pen has blown leads in only 24.6% of those starts, compared to an average of 28.3% over the past five years. That's partly because he's given them more than their share of big leads, and partly because we're using the past five years as our benchmark, but I think it's still fair to conclude that his bullpens haven't been quite as bad as they initially appeared.
• his overall ratio of blown leads (42) to off-hooks (15) is not good, but it's largely because he gives his bullpen the lead so often. The five year ratio of leads to deficits is about 50:50, but Clemens' ratio is 62:38.
• more than 60% of his wins (129 of 213) have required at least some assistance from his bullpen.

So Clemens has been a little unlucky over the years, but not nearly as much as I first thought. And it's clear that he's earned every bit of his .644 career winning percentage by completing more than a quarter of his starts and by giving his team the lead much more often than most.

Leaders Since 1993

Now let's take a look at some of the leaders during the past five years and in 1997. We'll start with the pitchers who have had their bullpens blow the most leads since 1993 (top 15 and ties are shown):

Jaime Navarro         18
Andy Benes            18
Wilson Alvarez        17
John Burkett          17
Ben McDonald          16
Mark Clark            16
Roger Clemens         16
Doug Drabek           16
Ricky Bones           16
Pete Harnisch         16
Ken Hill              16
Bob Tewksbury         15
Steve Avery           15
Andy Ashby            15
Scott Erickson        15
Mark Langston         15

Obviously, there's a bias in this list "in favor" of guys who made more starts and handed their bullpens more leads, so let's rank the guys who've seen leads blown in the highest percentage of their starts in the past five years (minimum of 60 starts to qualify):

Aaron Sele           .206
Ricky Bones          .203
Chris Haney          .194
Mark Clark           .186
Kirk Rueter          .183
Pedro Astacio        .176
Armando Reynoso      .174
Andy Ashby           .161
Kevin Foster         .159
Wilson Alvarez       .159
Allen Watson         .159
Curt Schilling       .157
Ben McDonald         .152
Kenny Rogers         .147
Donovan Osborne      .143

Note that Clemens doesn't show up on this list. In fact, in the past five years, 88 other starters have given their team a lead and seen it wasted in a higher percentage of their starts.

How about the lucky ones? The guys who've been bailed out by their teammates the most. Here are the top 15 and ties over the past five years:

Pedro Astacio         16
Kenny Rogers          16
Pedro Martinez        15
Joey Hamilton         15
Steve Avery           15
Jamie Moyer           14
Doug Drabek           14
Pat Rapp              14
Scott Erickson        14
Ron Darling           13
Orel Hershiser        13
Ken Hill              12
Dennis Martinez       12
Bill VanLandingham    12
Ramon Martinez        12
Roger Pavlik          12
Tom Candiotti         12
Kent Mercker          12
Mark Leiter           12
Steve Trachsel        12

And here are the pitchers who've rescued most frequently over the past five years, as a percentage of their starts (minimum of 60 starts):

Pedro Astacio          .216
Scott Kamieniecki      .183
Bill VanLandingham     .179
Kent Mercker           .176
Pat Rapp               .171
Kenny Rogers           .168
Joey Hamilton          .163
Pedro Martinez         .160
Roger Pavlik           .158
Kirk Rueter            .150
Chris Hammond          .145
Marvin Freeman         .133
Aaron Sele             .132
Steve Trachsel         .128
Esteban Loaiza         .123

Hmmm. A few really good pitchers who may show up here because they tend to leave behind smallish deficits. A bunch of mid-pack starters who are good enough to be given the required 60 starts but haven't ever been considered the aces of their staffs. I'm not sure what this list means, if anything, except that some of these guys might have W-L records a little better than they deserve and may therefore not be worth what they seem either in reality or fantasy ball.

The following starters were abandoned by their bullpens most often in 1997:

Mark Clark            7
Jeff Fassero          6
David Mlicki          6
Kirk Rueter           6
Steve Trachsel        6
Jose Rosado           6
John Thomson          5
Omar Olivares         5
John Burkett          5
Mike Mussina          5
Pat Hentgen           5
Dustin Hermanson      5
Curt Schilling        5
David Cone            5

It doesn't surprise me to see a lot of very good pitchers on this list. You have to give your team a lot of leads to have any chance of showing up here. Quite a few of them had good W/L records despite these blown leads (Clark 14-8, Fassero 16-9, and Rueter 13-6, for instance), while others (Mlicki and Trachsel, both 8-12) would have looked much more respectable had they received a little more help.

I wonder if Clark would have been traded by the Mets to Chicago if his luck had been better. Five of the blown leads occurred before the trade, and his pre-trade record could have been something like 12-7 instead of 8-7. True, his ERA with the Mets was 4.25, so a 12-7 record wasn't really justified. But with better bullpen support, he would have been near the league lead in wins at that point in the season, and how many clubs are going to trade a "winner"? After the trade, he was 6-1 for the Cubs despite two more blown leads.

Finally, here's a list of the pitchers who were taken off the hook most often in 1997:

Scott Kamieniecki     7
Jeff Suppan           6
Joey Hamilton         6
Mike Johnson          5
Jeff Juden            5
Pedro Astacio         5
Shane Reynolds        5
Steve Trachsel        4
Rick Reed             4
Pat Rapp              4
Scott Karl            4
Scott Erickson        4
Jason Dickson         4
Ariel Prieto          4

Showing up on this list shouldn't necessarily be considered a negative. It does mean that the pitcher left the game while behind quite a bit, but it doesn't mean he pitched badly. Consider the seven starts that earned Scott Kamieniecki the top ranking:

• 4/3 vs KC: 4 ER in 4-2/3 innings, left trailing 4-2 before O's rallied to win 6-4
• 4/9 at KC: left after 6 innings with a 2-0 deficit, Orioles won in 11
• 4/24 vs Bos: left a 1-0 game after seven, Sox won in extra innings
• 7/2 vs Phi: 4 ER in 5-2/3, was trailing 4-2, Philly ends up winning 10-6 when both bullpens blow up and trade the lead a couple of times
• 7/14 vs Tor: 3 ER in 5-1/3, saw his team rally from 3-2 down to win 9-5
• 8/29 vs Mets: came out with O's down 3-2 after seven, but Baltimore wins 4-3 in 12 innings
• 9/19 vs Det: left with a 3-0 deficit in the top of the 7th, O's rally for three in the ninth before losing in 11

While there are a couple of sub-par performances here, none were awful, and he kept his team in the game for the most part.

I guess that's enough for now. Please understand that this is not intended as a scientific study of starter or bullpen performance. There's a big difference between giving your team a one-run lead (a) with nobody out and two runners on in the 6th inning and (b) with nobody on and one out in the 9th. There's a big difference between handing your pen a 1-0 lead in the 8th and a 7-6 lead in the 6th. And a two-run lead means a lot less in Coors Field than in Dodger Stadium. So these numbers cannot, by themselves, be used to prove that one starter was more effective than another or that team X's bullpen was the worst over the past five years.

Even so, it was fun found to look at Clemens and other starters this way. Before doing this, I didn't have an intuitive sense of the average winning percentage in complete games, how often leads of various sizes are preserved by the bullpen as a whole (as opposed to focusing just on save percentages by closers), and which pitchers have been most and least helped by their pens in recent years.

And now we know that Clemens has been let down by his mates, but only a little, and not in the way I originally thought. It's true that they've blown quite a few leads, but only because he's given them so many in the first place. Their lack of support has been evident in their inability to fight back in many games after he's left them with a one- or two-run deficit.

Copyright © 1998. Diamond Mind, Inc. All rights reserved.