The World's #1 Starting Pitcher

September 18, 2011

The World’s Number One Starting Pitcher

 

            OK, here’s an obvious question:  why doesn’t baseball have an official ranking, updated once a week, as to who is the World’s #1 Starting Pitcher?

            Tennis and Golf have rankings of their players; the world’s #1 player, the world’s #2 player. . ..the world’s #150 player.    Tennis and Golf are individual sports, but you have to admit:  it’s a pretty cool concept.   Think about the role that this ranking would play in coverage of the sport, if it did exist.    First, there would be the battle for the number one spot.   At some point in their careers Steve Carlton and Tom Seaver would have been going head-to-head for the head of the class, as would Bob Gibson against Juan Marichal after Koufax retired, and Warren Spahn against Robin Roberts in the 1950s, and Dwight Gooden against Roger Clemens in mid-1980s, and Clemens against Maddux in the mid-1990s, and Tom Seaver against Jim Palmer when it was not Seaver against Carlton.

            If it did exist, the list would play into and would create a thousand other story lines within the sport.    Once or twice a year there is going to be a game where #1 faces off against #2, and that’s going to be a big event.   It’s a big event now, but. . .we don’t exactly know who is #1 and #2 now.    Any time two guys in the top ten square off against each other, it’s going to be an event.   Any time two guys in the top 25 square off, it’s going to be part of the story.  

            If the #143 pitcher beats the #1 pitcher—which is going to happen a couple of times a year—that’s a story.   When a young pitcher comes along and charges rapidly through the ranks, vaulting from 208th to 20th in two months, that’s going to be a part of his story.   When a former #1 gets knocked around and drops out of the top 100, that’s going to be a part of his story.  

            When a pitcher is a candidate for the Hall of Fame, the fact that he was the #1 ranked pitcher in the world for two years is going to be a major credential that his supporters can point to—or, vice versa, if the pitcher was never #1, it’s a major hole in his resume.   I would venture to say that Hal Newhouser would have gotten into the Hall of Fame years earlier if his supporters had been able to say that Newhouser spent almost four years ranked as the #1 pitcher in baseball—a fact that I suspect is true, although I haven’t done rankings from the 1940s.   Where a pitcher is on the list is going to be introduced in salary arbitration cases—for and against the player.    If a pitcher is #44 and trying to compare himself to a pitcher who is #8, it’s going to be mentioned.   

            What I am saying is that this ranking, if it did exist, would give shape and form to story lines that exist now, but which are vague and less interesting without precise dimensions.   The list, if it were published, would draw distinctions into a much sharper focus.   We know that Zack Greinke is not quite what he was two years ago, but how far exactly has he fallen?    Was he ever the #1 starter in the game?   Is he in the top 20 now?   The top 30?  The top 50?   We don’t exactly know, and because we don’t exactly know, we don’t know.

            The question will be ask "Why World #1 Starting Pitcher?  Couldn’t you do make the world’s #1 ranked second baseman, or left fielder, or relief pitcher?"

            You could, but it’s not the same.   The Starting Pitcher frames the debate for the game.    The starting pitcher’s name is in the paper.   Bettors bet based on the starting pitchers.    The starting pitchers are usually announced for several days in advance.

            There are about 150 starting pitchers in rotation at any moment.    It’s a "functional number", a critical mass.   You’ve got 30 second basemen; a lot of times the #1 guy isn’t going to change for two years.    Starting pitchers can elbow for position; they can move from 87th to 82nd based on one good start, up to 81st based on another, 78th based on a third, drop down to 84th based on a bad start.    There’s a line. 

            We say that one starting pitcher has beaten another, or has defeated another; it may not be the literal truth, but we say it all the time and we know what it means.   Nobody says that one second baseman has defeated another one, and no one would know what it meant if we did say it.  

            There’s a line.   Every team has a #1 starter, a #2 starter, a #3 starter, etc.   A team doesn’t have a #3 second baseman.    The concept works for starting pitchers in a way that it doesn’t work anywhere else in baseball.

            OK, you get the concept or you don’t; I don’t want to hype it, so let’s move on to the three practical questions.

            1)  How would we rank the pitchers?

            2)  Who have been the #1 ranked pitchers over the last ten or fifteen years?

            3)  Who is the #1 ranked pitcher now?

           

1)  How would we rank the pitchers?

            It’s pretty easy, actually.   I’ve played around with the formulas, re-constructing what the rankings would have been over the last twenty years, and I’m pretty much certain this is the way to do it.  

            Everybody starts out with a ranking of 300.000, and you can’t go lower than 300, even if you pitch badly.   If you’re at 300, you’re unranked; you’re only actually on the list if you have a current score higher than 300.  There would typically be 150 to 180 pitchers who are, at the moment, ranked.   Pitchers never actually pitch badly enough that they would rank below 300 (if it were possible to do so) for more than two or three starts, because if you pitch that badly, you lose your position in the rotation. 

            When a pitcher makes a start, we:

            a)  Mark down his previous ranking by 3%, and

            b)  Add 30% of his Game Score for the start. 

            We base the rankings on Game Scores, which means that we ignore wins and losses, but give weight to innings pitched, runs allowed, earned runs allowed, walks and strikeouts.  

            That’s really all there is to the system, except that there are nagging little problems the complicate the calculation while contributing very little, as there always are.    The three nagging little problems that we have to deal with are:

            a)  inactivity,

            b)  park effects, and

            c)  post season play.  

 

            For inactivity, these are the rules:

            1)  If a pitcher does not make a start for 1 to 6 days, his score does not change.   It remains whatever it was after his last start.

            2)  On days 7 to 200, if a pitcher does not make a start (for seven days or more), we reduce his ranking by one-quarter of a point for each day that he is inactive—in season or off season.   During the off-season everybody moves down, but everybody moves down in lock-step, so the rankings don’t change once you get 7 days from the end of the season.

            3)  If the pitcher remains inactive for more than 200 days, we reduce his score by one point per day beginning with the 201st day. 

           

 

            For park effects, you may remember that a few days ago I introduced a system to park-adjust Game Scores.    Take the average number of runs per game scored in the last 100 games played in the park; call that number R.   The expected Game Score in a game is:

 

            68 minus two times R

 

            which we will call "E" for "expected Game Score".

 

            E = 68 – 2R

 

            Then you adjust the Game Score by adding 50, and subtracting E.   GS is "Game Score" and AGS is "Adjusted Game Score":

 

            AGS = 50 + GS – E

 

            If a pitcher pitches in a hitters’ park this will increase his scores for his home games by perhaps 3 points a game, which makes a difference because it’s a consistent bias.   Ubaldo Jimenez, when he was a great pitcher a year or two ago, was making start after start after start in Coors’ Field.   It creates a problem in the rankings if you ignore that.   Some of you will think I should have adjusted for the quality of the competition faced in the start, but actually, that’s not necessary and not helpful.   That’s a transient bias; it changes all the time.    Adjusting for a transient bias in a constantly fluctuating system is just a waste of time and energy, sort of the sabermetric equivalent of opening your refrigerator door on a hot day to feel the little blast of cool air.   

 

            That’s all there is to the system.   Oh. . .post-season play.   Yes, we include post-season starts, and include them on exactly the same basis as regular season starts.  

            Yeah, I know there’s going to be a fear here that this will cause the system to discriminate against Felix Hernandez or Clayton Kershaw or somebody, but. . .it’s really not a meaningful issue.   Including the post-season starts gives the pitcher on a good team two potential advantages.   If he pitches well, his score goes up—but also, because he is delaying the off-season, he is putting off by a few days the point at which his score begins to slide backward, thus giving the pitcher an advantage.

            An advantage of maybe two points, which is not very meaningful.   Also, what you have probably never realized is this:  most pitchers who make it into post-season play don’t end their season with a good start.    You know how, in the NCAA basketball tournament, every team except one ends the season with a loss?   For the same reason, when a pitcher’s team goes into post season, most pitchers wind up the season with a bad outing.   Thus, for those pitchers—the great majority of them—it’s really not an advantage to be in post-season play; it’s an advantage and a disadvantage, and they wash out at best.   A bad start knocks a pitcher backward much further than a few days inactivity.   It’s only an advantage for the pitcher who gets into post-season play and pitches several good games—and you just can’t ignore those games; we can’t say that what Roy Halladay has done in post-season play is irrelevant to where he ranks among the game’s best pitchers.    You can think what you want, but I’m certain that that is the right answer.  

 

            A couple of philosophical questions here, which have no philosophical answers, only practical ones. 

            1)  Why do we use Game Scores?, and

            2)  Why do we discount the old score by 3% per outing, as opposed to 6% or 8%?

            We use Game Scores because what else are you going to use, if you’re going to be constantly updating the list?   We want to re-evaluate the pitcher’s standing every time he pitches, right?  What are you going to use, other than Game Scores? 

            Whether 3% is the right number or something else is a judgment call.   Suppose we start at 10%.  

            If you replace 10% of the pitcher’s "score" after each start, then 69% of the pitcher’s score will be based on what he has done in the last two months (assuming 11 starts per two months), 90% will be based on what he has done in the last four months, and 97% will be based on what he has done in the last year.    That becomes a "hot pitcher" list rather than a "great pitcher list".   What we’re looking for is the deepest level of confidence about a pitcher’s ability.   Clayton Kershaw may have had a better year this year than Roy Halladay, but Roy Halladay is still a better pitcher than Clayton Kershaw based on what he has done over a longer period of time.

            Not dramatically better; when a pitcher sustains excellence for a full season, that’s a meaningful time period, but obviously 10% is too high a percentage to be put onto one start.

            Try 1%.   If each start was worth only 1% of the pitcher’s ranking, then a full season’s starts would only make up 28% of the pitcher’s ranking.   72% of the ranking would be based on what the pitcher did before the last year.   Clayton Kershaw would still be nowhere near the top of the list.   Obviously, that’s not right. 

            Three percent is the ratio that works.   At 3%, 28% of the pitcher’s rating is based on what he has done in the last two months, 49% based on the last four months, 63% based on the last full season.    That’s about the right ratio.   MOST of the evaluation of a pitcher—over half—has to be based on what he has done in the last year.  

 

 

 

2)  Who Have Been the #1 Ranked Pitchers over the Last Fifteen to Twenty Years?

 

            Clayton Kershaw has come up here a couple of times, so let’s use him to illustrate the process.  

            Kershaw’s first major league start was on May 25, 2008, against St. Louis in Dodger Stadium.   Kershaw pitched well, posting a box score line of  6    5  2  2  1  7, which is a Game Score of 60.    The average runs scored per game in Dodger Stadium in 2008 were 7.60 (3.80 per team), so the expected game score for a starter was 52.80.     This is over 50—pitcher’s park—so that changes Kershaw’s Game Score from 60 into an adjusted Game Score of 57.20. 

            Kershaw’s "previous ranking" was 300.00, which is the default rank for a pitcher who has never pitched before.   We discount that by 3%, marking it down to 291.00, and then add 30% of the Adjusted Game Score (57.20).  30% of 57.20 is 17.16, so we add 17.16, and Kershaw’s Ranking, after his first start, is 308.16.

            Following that start, Kershaw had a series of not-very-good starts; through nine major league starts his ERA was 5.18.    These were his Game Scores for those next eight games:

 

May

30

33

June

4

50

June

10

49

June

15

58

June

20

45

June

26

41

July

1

47

July

22

18

           

            Only one game of the eight scored above 50.   However, because Kershaw was ranked barely over 300, these low scores were enough to push him gently forward—not nearly as fast as if he had pitched well, certainly, but he was just working his way into the system:

 

 

 

Game

New

Month

Day

Score

Rating

May

30

33

308.87

June

4

50

313.77

June

10

49

318.25

June

15

58

327.15

June

20

45

330.00

June

26

41

331.56

July

1

47

335.84

 

            His score went up after every start.   After the July 1 start, however, two things happened:

            1)  Kershaw didn’t start again for 21 days, and

            2)  His next start was really bad.  He gave up 10 hits and 5 runs in three innings, posting a Game Score of 18.    His score was marked down by 3.75 points because he was inactive for 21 days, and it was knocked down more because his Game Score was only 18:

 

 

 

Game

New

Month

Day

Score

Rating

June

26

41

331.56

July

1

47

335.84

July

22

 

321.88

 

            The combination of the layoff and the bad start cost him fourteen points in the ranking, 40% of what he had gained up to that point.

            Following that, however, Kershaw had a series of very good starts, with Game Scores ranging from 54 to 70.   Now Kershaw rose more rapidly:

 

 

Game

New

Month

Day

Score

Ranking

July

27

68

337.99

August

1

66

346.82

August

7

70

357.42

August

12

54

362.06

August

17

57

367.46

 

            After the August 17 start Kershaw’s ERA was 3.59.   By the end of the year it was 4.26.   Kershaw generally did not pitch well over the last six weeks of the season, but still, Kershaw’s ranking was only 367, which means (essentially) that his ranking went up if he had a Game Score of 37 or better.    By the end of his first full season—a so-so season, 20 starts, 4.26 ERA—Kershaw’s Ranking Score was 388.33.

            Then there was the off season.   Kershaw’s last start in 2008 was on September 24; his first start in 2009 was April 9.  That’s a gap of 197 days, which creates a markdown of 47.75 points.   Thus, Kershaw’s ranking, before his start of April 9, 2009, was 340.58.   He started out 2009 forty points ahead of where he had started out 2008. 

            A pitcher starts at 300, and an average Game Score is about 50, so the pitcher is pushing himself toward 500.   However, only the very good pitchers actually reach 500; it’s not that easy to get there.    The inactivity rules act as gravity, pulling the pitchers’ scores back toward 300.  Essentially, only the #1 starters ever get ranking scores as high as 500.  

            Kershaw pitched much better in 2009 than he had in 2008, 30 starts with an ERA of 2.79.    By the end of the regular season in 2009 his score was up to 478.95.   He pitched twice in the post-season in 2009.   He pitched OK the first time and was hit hard the second time, leaving him, after his second post-season start, at 473.00.  

            Gravity pulled him backward again over the off-season, this time by 40.5 points, and Kershaw began the 2010 season at 432.50.   (Illustrating the point made before, if Kershaw had not pitched in the post-season in 2009, his discount for the off-season would have been larger, but his score would have been 433.95—essentially the same.)    In any case, Kershaw had another fine season in 2010, making 32 starts with a 2.91 ERA.   By the end of the 2010 season his score was in #1 starter territory, at 527.63.  Kershaw ranked as the #23 starting pitcher in the major leagues at the conclusion of the 2010 regular season, between Dan Haren and Zack Greinke.   

            Over the off-season Kershaw dropped backward by another 46 points, as all the pitchers did, starting the 2011 season at 481.88.    This season has been brilliant, and Kershaw currently ranks (September 18) at 582.02.    He is currently the #5 ranked starting pitcher in the world, behind. . . .well, I’m getting a little ahead of myself.

 

 

 

2)  Who Have Been the #1 Ranked Pitchers over the Last Fifteen to Twenty Years?

 

            Which I promised you before.  

            OK, the problem is where to start.   The only really accurate place to start the rankings is would be in 1876, but of course I don’t have organized data for every start prior to 1960.    I started my rankings with the 1990 season.

            As best I know, the #1 starting pitcher in the world, two months into the 1990 season, was Roger Clemens.   It is theoretically possible, because I am not using data before 1990, that that is not the right answer.   Roger Clemens had pitched really well for several years prior to 1990, and I would guess that he might have been the #1 ranked starting pitcher even before 1990, but that’s a guess; coming into the 1990 season the #1 guy could have been Clemens, Saberhagen or Orel Hershiser.

            Hershiser was hurt early in the 1990 season; Saberhagen could have (and probably did) run neck-and-neck with Clemens early in the 1990 season, but then he also was injured.   Clemens had another brilliant season (21-6, 1.93 ERA.)  By June 1, 1990, I am assuming that Roger Clemens was the #1 ranked starting pitcher in baseball, and would have held that position until very late in the season.

            In very late September, however, Roger Clemens would have lost that position to Dave Stewart of the Oakland A’s.   Stewart—a 20-game winner in 1987, 1988 and 1989—would have entered 1990 in a very strong position, among the five top starting pitchers in baseball, and he pitched 267 innings in 1990 with an ERA of 2.56.   Clemens had an injury of some sort, pitched badly on September 4 (1990) and then was inactive for 21 days, pitching again on September 25.   Stewart continued to pitch brilliantly, pitching 47 innings in September, 1990 with a 1.47 ERA (actually includes one start on October 1), and this would have put Stewart in the #1 spot at season’s end.   (The 1990 American League Cy Young Award winner, Bob Welch, would not have ranked in the top ten.  Which is quite unusual, as we will see.)

            Very, very early in the 1991 season, Clemens would have regained his #1 ranking, and would have held that position for the rest of the 1991 season without any serious challenge.   As best I know, these would have been the standings at the end of the 1991 regular season:

 

First

Last

Score

Rank

Roger

Clemens

570.87

1

Nolan

Ryan

541.06

2

David

Cone

540.65

3

Jose

Rijo

531.91

4

Mark

Langston

524.82

5

Greg

Maddux

521.57

6

Jack

McDowell

520.14

7

Andy

Benes

519.10

8

Tom

Glavine

515.91

9

Doug

Drabek

510.97

10

\

            Nolan Ryan by this time was 44 years old.   He pitched well, though—a 2.91 ERA with 203 strikeouts in 173 innings—and he had gone 9-2 over the last three months of the season after starting out 3-4.   Jack Morris ended the 1991 regular season in 12th place.   He made five starts in the 1991 post-season, however, the last four of them good and the last one brilliant, the famous ten-inning shutout, Game Score 84.  That would have vaulted him into 7th place at the beginning of the 1992 season.  

            Clemens, however, continued to roll, and continued to dominate through most of the 1992 season.    His closest competitor the first half of the 1992 campaign was David Cone, trailing him generally by about 30 points.   In early July Greg Maddux moved into second place, and over the second half of the season Maddux closed steadily on Clemens.   By early September Maddux trailed Clemens by only about ten points.    On September 12, 1992, Clemens was hit hard by Detroit (7 hits, 5 runs in 6 innings), and on September 17 he was shelled by Milwaukee (10 hits, 8 runs in four and a-third.)    Greg Maddux—not pitching sensationally either, at the moment-- became the #1 starting pitcher in baseball. 

              Maddux held the #1 spot for the rest of the 1992 season and for the first three weeks of the 1993 season.   On April 25, 1993, however, Maddux lasted only four innings against St. Louis (9 hits, 4 runs), while Clemens pitched extremely well against the Angels—losing the game 2-1, but in 8 innings he gave up 4 hits, 2 runs, struck out 9 and walked nobody.   Clemens moved back into the #1 spot:

 

            June 1, 1990 to September 13, 1990               Roger Clemens

            September 14, 1990 to April 8, 1991              Dave Stewart

            April 9, 1991 to September 16, 1992              Roger Clemens

            September 17, 1992 to April 24, 1993            Greg Maddux

            April 25, 1993                                           &​nbsp;     Roger Clemens

 

            Clemens pitched well early in 1993, but finished the season with a 4.46 ERA.   Maddux moved back into the number one spot on May 21, and within a few weeks after that, Clemens was no longer his chief competitor.  

            In July and August, 1993, Jose Rijo pitched a series of extremely good games, and surged briefly into the number one spot.   Mostly forgotten today, Jose Rijo was a really good pitcher.  He posted ERAs of 2.39 in 1988, 2.84 in 1989, 2.70 in 1990, 2.51 in 1991, 2.56 in 1993 and 2.48 in 1994.   In the 1990 post-season he was 3-0 with a 2.28 ERA.    He had generally been one of the top five pitchers in baseball, throughout that era.   From July 11 to August 25, 1993, Rijo made ten starts, pitched 7 or 8 innings in every start, and allowed only 9 earned runs, a 1.20 ERA for a third of a season’s work.  He moved to the top of the list.

            On August 15, 1993, Maddux matched up against Rijo in Cincinnati.   Rijo was really good, giving up only 4 hits, one run, striking out 11.   Maddux was better; Maddux pitched a 4-hit shutout.   Maddux won the game, 1-0—and moved back into the #1 spot:

 

June 1, 1990 to September 13, 1990               Roger Clemens

            September 14, 1990 to April 8, 1991              Dave Stewart

            April 9, 1991 to September 16, 1992              Roger Clemens

            September 17, 1992 to April 24, 1993            Greg Maddux

            April 25, 1993 to May 20, 1993                      Roger Clemens

            May 21, 1993 to August 4, 1993                    Greg Maddux

            August 5, 1993 to August 15, 1993               Jose Rijo

            August 15, 1993                            ​                 Greg Maddux

 

            By May, 1994, Greg Maddux was 50 points ahead of the field.   In 1994 Maddux was "just" 16-6, but with a 1.56 ERA.  In 1995 he was 19-2 with a 1.63 ERA.    On June 22, 1994 he hit 600—a historic level.   By the time of the strike in 1994 he was at 626.    The long off-season (because of the strike) pulled him backward further than normal, but by July 1, 1995, he was back at the 600 level, still 40 points ahead of his closest competitor.   By mid-August, 1995 he was at 627, and Sports Illustrated put him on the cover with the headline "The Greatest Pitcher You’ll Ever See."  By the end of the 1995 season he had peaked at 637.99.   

            But by the end of the 1995 season, Randy Johnson was also over 600.   Maddux was 19-2 in 1995; The Big Unit was 18-2 with 294 strikeouts in 214 innings.   He wasn’t chopped liver, either.     Maddux wasn’t quite as good in 1996, but Johnson was injured, and by mid-season, 1996, Maddux was sixty points ahead of his closest competitor, who was his teammate, John Smoltz.  Smoltz won 24 games in 1996 and the Cy Young Award (Pat Hentgen won in the American League), and by the end of the 1996 season these were the standings:

 

First

Last

Score

Rank

Greg

Maddux

592.92

1

John

Smoltz

571.01

2

Kevin

Brown

560.40

3

Hideo

Nomo

546.22

4

Roger

Clemens

533.09

5

Kevin

Appier

529.41

6

Alex

Fernandez

527.60

7

Pedro

Martinez

520.64

8

Pat

Hentgen

519.42

9

Tom

Glavine

516.42

10

 

            Maddux had now held the #1 spot for more than three full seasons.  

            Greg Maddux in 1997 was 19-4 with a 2.27 ERA—not really a bad season.   Roger Clemens, however, opened the 1997 season 11-0 with a 1.69 ERA—and he had ranked as the #5 pitcher in the game coming into the season.    On July 12, 1997, Clemens, now pitching for Toronto, returned to Fenway Park.  He struck out 16 batters and walked no one, posting a Game Score of 86.    For the first time in almost four years, Greg Maddux was no longer the #1 starting pitcher in baseball:

 

June 1, 1990 to September 13, 1990               Roger Clemens

            September 14, 1990 to April 8, 1991              Dave Stewart

            April 9, 1991 to September 16, 1992              Roger Clemens

            September 17, 1992 to April 24, 1993            Greg Maddux

            April 25, 1993 to May 20, 1993                      Roger Clemens

            May 21, 1993 to August 4, 1993                    Greg Maddux

            August 5, 1993 to August 15, 1993               Jose Rijo

            August 15, 1993 to July 11, 1997                   Greg Maddux

            July 12, 1997                           &n​bsp;                       Roger Clemens

 

            In the last two decades, no one else remotely approaches Greg Maddux’ four-year death grip on the #1 spot.

            By July 28, 1997, Clemens was at the historic 600 level.   In mid-August, 1997, he was joined at that level by Pedro Martinez.   On August 25, 1997, Pedro pitched 8 and two-thirds innings against St. Louis, striking out 13 and allowing only 4 hits, one run, and cutting his ERA for the season to 1.61.    He moved into first place.

            For a few weeks there there was a fantastic three-man duel for first place, between Clemens, Pedro and Maddux, all of them hanging in the range of 595 to 610 points.   By May, 1998, however—the scores of course had been dragged backward by the off season—Pedro had pulled about 20 points ahead, and in second place at that time was Curt Schilling.  

            Pedro was traded to the Red Sox (for Carl Pavano and Tony Armas) on November 18, 1997.   On May 20, 1998, Pedro had an ERA for the season of 1.74.   He had a series of rough outings then, however, and by June 10 his ERA was 3.46.   On June 12, 1998, Greg Maddux moved back into first place:

 

 

June 1, 1990 to September 13, 1990               Roger Clemens

            September 14, 1990 to April 8, 1991              Dave Stewart

            April 9, 1991 to September 16, 1992              Roger Clemens

            September 17, 1992 to April 24, 1993            Greg Maddux

            April 25, 1993 to May 20, 1993                      Roger Clemens

            May 21, 1993 to August 4, 1993                    Greg Maddux

            August 5, 1993 to August 15, 1993               Jose Rijo

            August 15, 1993 to July 11, 1997                   Greg Maddux

            July 12, 1997 to August 24, 1997                   Roger Clemens

            August 25, 1997 to June 11, 1998                  Pedro Martinez

            June 12, 1998                             ​                     Greg Maddux

 

            It was the fourth time that Maddux had been ranked as the world’s #1 starting pitcher.   Maddux held the position for more than two months, which actually is a long time, albeit not compared to four years.   On August 23, 1998, however, Maddux gave up 10 hits and 7 runs to the Dodgers, lasting just 5 innings; his ERA ballooned from 1.65 to 1.91.  On August 25, Roger Clemens pitched a three-hit shutout against Kansas City, striking out 18 and walking no one.   The Rocket was back in first.   He was in the process of winning his fifth Cy Young Award, and his second in a row.

            Clemens ruled this time until April 27, 1999.   On that date—now pitching for the Yankees--he was knocked out after two innings in Texas.    That put Randy Johnson in first place—for four days:

 

June 1, 1990 to September 13, 1990               Roger Clemens

            September 14, 1990 to April 8, 1991              Dave Stewart

            April 9, 1991 to September 16, 1992              Roger Clemens

            September 17, 1992 to April 24, 1993            Greg Maddux

            April 25, 1993 to May 20, 1993                      Roger Clemens

            May 21, 1993 to August 4, 1993                    Greg Maddux

            August 5, 1993 to August 15, 1993               Jose Rijo

            August 15, 1993 to July 11, 1997                   Greg Maddux

            July 12, 1997 to August 24, 1997                   Roger Clemens

            August 25, 1997 to June 11, 1998                  Pedro Martinez

            June 12, 1998  to August 24, 1998                 Greg Maddux

            August 25, 1998 to April 26, 1999                 Roger Clemens

            April 27, 1999 to April 30, 1999                    Randy Johnson

            May 1, 1999                        &nb​sp;                        &nb​sp;  Pedro Martinez

 

            Johnson, who moved into first place by default when Clemens was cuffed around, made one start as the #1 pitcher and pitched very well, but had only a razor thin lead over Pedro Martinez.   When Martinez started he pitched even better, and he got to hold the brass ring.

            Pedro was having a historic season, finishing 23-4 with 313 strikeouts in 213 innings—and yet he was never more than a few points ahead of Randy Johnson.    On July 18, 1999, Pedro had one of the worst games of his career, giving up 12 hits and 9 runs in three and two-thirds innings against the Marlins.  (Pedro had not pitched in ten days before that game, obviously suggesting that there was some sort of an injury involved.)  In any case, that put Randy back in first place.

            Randy Johnson struck out 364 batters in 1999, leading the National League in ERA at 2.64.   Pedro Martinez, despite his historic season, spent the second half of that season and the first half of 2000 as baseball’s #2 ranked starting pitcher.    

            On June 8, 2000, Pedro Martinez defeated Cleveland 3-0, giving up only 1 hit in 8 innings while striking out 10.   His ERA for the season, after eleven starts, was 0.95:

 

June 1, 1990 to September 13, 1990               Roger Clemens

            September 14, 1990 to April 8, 1991              Dave Stewart

            April 9, 1991 to September 16, 1992              Roger Clemens

            September 17, 1992 to April 24, 1993            Greg Maddux

            April 25, 1993 to May 20, 1993                      Roger Clemens

            May 21, 1993 to August 4, 1993                    Greg Maddux

            August 5, 1993 to August 15, 1993               Jose Rijo

            August 15, 1993 to July 11, 1997                   Greg Maddux

            July 12, 1997 to August 24, 1997                   Roger Clemens

            August 25, 1997 to June 11, 1998                  Pedro Martinez

            June 12, 1998  to August 24, 1998                 Greg Maddux

            August 25, 1998 to April 26, 1999                 Roger Clemens

            April 27, 1999 to April 30, 1999                    Randy Johnson

            May 1, 1999 to July 17, 1999                         Pedro Martinez

            July 18, 1999 to June 7, 2000                         Randy Johnson

            July 8, 2000                                          ​;           Pedro Martinez

 

            It was the third time that Pedro had been in the #1 spot. 

            Among the themes that I am trying to convey here is that sometimes the lead changes because a pitcher had a couple of bad games.   Other times, the lead changes because somebody else just pitches at a fantastic level.   Pedro in 2000 was pitching at a fantastic level, finishing the season "only" 18-6 but with a 1.74 ERA, 284 strikeouts in 213 innings.   But he pitched only 213 innings; he had some little outages there when he would miss a start.   When he missed a start, the Big Unit was ready.

            Over the next six years (1999-2005) Pedro and the Unit engaged in a quite remarkable duel for the #1 spot.  

 

April 27, 1999 to April 30, 1999                    Randy Johnson

May 1, 1999 to July 17, 1999                         Pedro Martinez

            July 18, 1999 to June 7, 2000                         Randy Johnson

            June 8, 2000 to June 28, 2000                        Pedro Martinez

            June 29, 2000  to July 22, 2000                      Randy Johnson

            July 23, 2000 to July 28, 2001                        Pedro Martinez

            July 29, 2001 to July 24, 2002                        Randy Johnson

            July 25, 2002 to August 24, 2002                   Pedro Martinez

            August 25, 2002 to May 2, 2003                    Randy Johnson

            May 3, 2003 to July 16, 2004                         Pedro Martinez

            July 17, 2004 to July 29, 2004                        Jason Schmidt

            July 30, 2004 to August 11, 2004                   Randy Johnson

            August 12, 2004 to August 30, 2004             Jason Schmidt

            August 31, 2004 to September 13, 2004        Randy Johnson

            September 14, 2004                        &nbs​p;               Pedro Martinez

            September 15 2004 to May 26, 2005              Randy Johnson

            May 27, 2005  to August 12, 2005                 Pedro Martinez

 

            Pedro’s score, as of June 20, 2000, was 650.64.   Johnson’s score on June 29 was 650.24.   Six hundred is a historic level.   Both Pedro and Johnson, at that time, were pitching at nearly unprecedented levels of effectiveness, challenging if not defeating Sports Illustrated’s assertion that we would never see the equal of Greg Maddux (although it should be noted that Maddux’ peak score, 639, was kicked backward by the 1994-1995 strike, which created a long off-season.)   By the end of the 2000 season, anyway, Pedro’s ranking score was 681.55.   This is the highest level attained by any pitcher over the last twenty years.   

And Johnson, throughout all of this era, was right with him or ahead of him.   They both had injuries.  Pedro had a serious injury in 2001, and didn’t win a game that year after May 30.   Johnson was hurt and out in 2003, finishing that season 6-8 with a 4.26 ERA.   After the injuries they were not the same, not the super-dominant pitchers that they had been from 1999 to 2002—but they remained the best in the game, with a brief interruption from Jason Schmidt, for almost three years beyond their peak.   Johnson in 2004 was just 16-14, but with a strikeout/walk ratio of 290 to 44—still perhaps the best pitcher in the game at age 40.

In 2004 Roger Clemens signed with the Astros, winning his sixth Cy Young Award there in 2004.   In 2005 he cut his ERA by more than a run.   On August 13, 2005, Clemens pitched eight innings of 2-hit, shutout ball, striking out 9 and walking no one--and returned to the #1 spot after an absence of six years:

 

            September 15 2004 to May 26, 2005              Randy Johnson

            May 27, 2005  to August 12, 2005                 Pedro Martinez

August 13, 2005                        ​;                     Roger Clemens

 

Pedro took it back again, lost it again:

 

            September 15 2004 to May 26, 2005              Randy Johnson

            May 27, 2005  to August 12, 2005                 Pedro Martinez

August 13, 2005 to August 24, 2005             Roger Clemens

August 25, 2005 to August 30, 2005             Pedro Martinez

August 31, 2005 to September 2, 2005          Roger Clemens

 

Clemens pitched on September 3, 2005, and pitched well—but lost his top spot anyone to a new power:

 

September 15 2004 to May 26, 2005              Randy Johnson

            May 27, 2005  to August 12, 2005                 Pedro Martinez

August 13, 2005 to August 24, 2005             Roger Clemens

August 25, 2005 to August 30, 2005             Pedro Martinez

August 31, 2005 to September 2, 2005          Roger Clemens

September 3, 2005                                 ​         Johan Santana

 

Johan Santana in 2004 was 20-6, 2.61 ERA—probably the best pitcher in baseball that year.   He had little history before that season, however, and our system demands a history; it demands that the pitcher pitch at a high level for two years, generally speaking, before it will consider him for the top spot.   By September, 2005, Santana had met that standard.   He would not surrender his position until May of 2006:

 

September 15 2004 to May 26, 2005              Randy Johnson

            May 27, 2005  to August 12, 2005                 Pedro Martinez

August 13, 2005 to August 24, 2005             Roger Clemens

August 25, 2005 to August 30, 2005             Pedro Martinez

August 31, 2005 to September 2, 2005          Roger Clemens

September 3, 2005 to May 25, 2006               Johan Santana

May 26, 2006                                     ​;             Pedro Martinez

 

By 2006 Pedro was in his second year with the New York Mets, and he was a ghost of the pitcher he had once been.   Had their been a Randy Johnson in his prime, a Pedro Martinez in his prime, a Roger Clemens or Greg Maddux, Pedro would not have been a serious candidate.   Santana had not yet reached that level.   Pedro in 2005 was 15-8, 2.82 ERA, strikeout to walk ratio of 208 to 47.   Through May of 2006 he was 5-1, 2.50 ERA.   It was the tenth and final time that he would hold the #1 spot.   He held for a little more than a week:

 

September 15 2004 to May 26, 2005              Randy Johnson

            May 27, 2005  to August 12, 2005                 Pedro Martinez

August 13, 2005 to August 24, 2005             Roger Clemens

August 25, 2005 to August 30, 2005             Pedro Martinez

August 31, 2005 to September 2, 2005          Roger Clemens

September 3, 2005 to May 25, 2006               Johan Santana

May 26, 2006  to June 7, 2006                        Pedro Martinez

June 8, 2006                                          &nbs​p;         Johan Santana

 

Santana is not unlike Pedro—a small pitcher who combines a good fastball with a devastating change.    He had no Big Unit to challenge him, and Santana owned the title now for more than two years—the longest that anyone had held the position since Greg Maddux held it for four years in the mid-1990s.  

On July 7, 2008, CC Sabathia was traded to the Milwaukee Brewers.    Sabathia—always a great pitcher—ripped off a string of stunning games, which put him into the lead chair:

 

September 3, 2005 to May 25, 2006               Johan Santana

May 26, 2006  to June 7, 2006                        Pedro Martinez

June 8, 2006 to July 22, 2008                         Johan Santana

July 23, 2008 to April 17, 2009                      CC Sabathia

April 18, 2009 to June 5, 2008                        Johan Santana

June 6, 2009                                    ​                CC Sabathia

 

 

Did I say that Santana had no Big Unit to challenge him?   They don’t get much bigger than Carsten Charlie.   In July, 2009, CC was briefly dislodged by Dan Haren, then by Tim Lincecum.  Since then we have been playing Hot Potato with it:

 

June 8, 2006 to July 22, 2008                         Johan Santana

July 23, 2008 to April 17, 2009                       CC Sabathia

April 18, 2009 to June 5, 2008                        Johan Santana

June 6, 2009 to July 9, 2009                           CC Sabathia

July 10, 2009 to July 31, 2009                         Dan Haren

August 1, 2009 to September 6, 2009            Tim Lincecum

September 7, 2009 to September 13, 2009     CC Sabathia

September 14, 2009 to September 18, 2009   Tim Lincecum

September 19, 2009 to September 30, 2009   CC Sabathia

October 1, 2009 to April 15, 2010                  Tim Lincecum

April 16, 2010 to April 19, 2010                     CC Sabathia

April 20, 2010 to May 3, 2010                        Roy Halladay

May 4, 2010 to May 17, 2010                         Tim Lincecum

May 18, 2010  to September 22, 2010            Roy Halladay

September 23, 2010 to October 5, 2010         Felix Hernandez

October 6, 2010 to July 30, 2011                   Roy Halladay

July 31, 2011 to the present                          Justin Verlander

 

Well. . .we’re played a little hot potato with it, but Roy Halladay held the hot potato, except for a couple of weeks here and a couple there, from April, 2010 into July, 2011.   That was a good run.

            Verlander took the lead on July 31, in that memorable contest against Jered Weaver.   I am not certain this is right, because (as I said) I don’t have highly organized data for 2011, but I BELIEVE that the current standings may be:

 

            1.  Justin Verlander

            2.  Roy Halladay

            3.  CC Sabathia

            4.  Cliff Lee

            5.  Clayton Kershaw

6.  Tim Lincecum

7.  Jered Weaver

8.  Felix Hernandez

9.  Matt Cain

10.  Cole Hamels

 

The World’s #1 Starting Pitcher, at this point, is Justin Verlander.   Zack Greinke, just to punch a button I left open earlier, is still in the top 20. 

 

 

 

 
 

COMMENTS (28 Comments, most recent shown first)

jdw
Fun stuff.

One thing that some might not like about this is how much it's bounced around. I'd point them to the ATP and WTA rankings:

http://en.wikipedia.org/wiki/List_of_ATP_number_1_ranked_singles_players

http://en.wikipedia.org/wiki/List_of_WTA_number_1_ranked_players

The WTA had 14 changes from 07/10/78 to 06/14/82 at Chris and Martina (and even Austin) traded the #1 spot before Martina went off on her record run.

The ATP has 33 changes from 04/09/79 to 09/09/85 between Connors dominance (245 out of 246 weeks) ending and Lendl's beinging (237 out of 257 only interupted by Willander winning three slams in a year). While there were some stretches in that period where a player was generally dominant, they still got broken up:

Borg: 100 out of 104 weeks

That stretch had two cups of coffer by Mac in it.

Mac: 164 out of 214 weeks

Mac had two full years (52+ weeks) at #1, but this strech was still split up by no less than 12 cups of coffee by Connors & Lendl.

That period (Connors-Borg-Mac + Martina-Chris) was the modern golden age of popularity of tennis in this county. We didn't pay as much attention to the rankings as people would today, with most of the focus being on the Year End Ranking. But if people didn't have an issue with the mens #1 spot bouncing around 33 times in 6.5 years, then they should have one with this.

Why people might have issue with it bouncing is because the era of Tiger and Roger having epic long runs at #1 in golf and mens tennis makes them think there always is a cast iron #1, and only passing of the torches.​
6:09 PM Sep 26th
 
bjames
Responding to Tango. . .and I should have read more carefully. I see that I just didn't read your previous comment as carefully as I should have.
11:47 PM Sep 22nd
 
chrisnickell
Great article. I really enjoyed the setup for this and the results. I'd love to see this become a regular feature - not just of the site but of baseball. How great would it be to see these rankings next to the pitcher match-up graphic - like we see college team ranks on their game graphics. Exactly as Bill says, a #8 vs. #22 looks a lot more exciting to me with the numbers than without them.

I like game score for this. I vastly prefer this to WAR or any other more complex measurement. Simplicity has great value. If someone wants a more complicated version they are welcome to go make it but I'd like something that I can explain to regular fans easily. This absolutely is that. Sure, there are lots of rules and I couldn't explain it all easily but anyone can get this. That is cool.

The only thing that bothers me is the DH. It is probably a minor factor - maybe it is even fully accounted for by the home field adjustment - but I don't like that National League pitchers would seem to have an advantage. Or do the fact that sometimes NL pitchers don't get to pitch quite as deep (because they are pinch hit for) balance this out?

Regardless, I think this is very cool. I'd certainly like to see it going back farther and would love to see it enter the mainstream.

Thanks,
Chris
4:01 PM Sep 22nd
 
tangotiger
I should have been clearer. I was responding to your point here:

"The second problem with the exponential decay process is that it leaves people in the system forever after they drop out of the rotation."

So, my counter to that is that a solution to the problem you noted could be solved easily enough, and that readers here could offer good solutions.

1:44 PM Sep 22nd
 
bjames

As for the second problem, you can always force a dropout after say two years of inactivity. In any case, I think a good solution can be implemented to address this problem. Plenty of smart people on this board can offer their insights.


The problem being what? I'm afraid I don't see a problem that needs to be solved.
9:35 PM Sep 21st
 
tangotiger
Bill,

The first problem is easily addressed, by changing it to "weeksAgo". Instead of .998^daysAgo, it's .998^(weeksAgo*7).

A pitcher who pitched within one week, would have weeksAgo=0, thereby setting the exponent to 0, and the weight = 1. That is, .998^0 = 1.

As for the second problem, you can always force a dropout after say two years of inactivity. In any case, I think a good solution can be implemented to address this problem. Plenty of smart people on this board can offer their insights.




3:24 PM Sep 21st
 
bjames
The first problem with the exponential decay system is that it doesn't represent the real world as well. The exponential decay system gives the pitcher a different score if he pitched his last game on Sunday than if he pitched on Friday. There should NOT be such a difference. The fact that a pitcher made his last start four days ago, rather than one, should not play any role in determining where the pitcher ranks--unless the pitcher hasn't started in a week, in which case it should.

The second problem with the exponential decay process is that it leaves people in the system forever after they drop out of the rotation. I experimented with an exponential decay in developing the system, but. . .it just doesn't really work well in practice.
2:52 PM Sep 21st
 
bjames


Responding to the issue of why the system starts a pitcher out at 300, rather than zero. . ..it is because scores under 300 would not be legitimate rankings for any pitcher, so you would be dealing with illegitimate evaluations of all of those pitchers, and this would cause massive, massive, massive problems in the rankings. The real issue isn’t whether we should start pitchers out at zero or 300; it is whether we should start them out at 300 or 350 or possibly 425.

The pitcher’s “legitimate” skill ranking is, in essence, ten times his average Game Score. ..we could say ten times his average recent Game Score, except that we don’t want to base it merely on what the pitcher has done recently; we want to be sure the pitcher is really as good as his score indicates, and not merely that he has had some good games lately.

To say that a pitcher has a rankings score of 190 would be a way of saying that his Game Score average is 19. Well, nobody is that bad; nobody is anywhere NEAR that bad. A. J. Burnett and John Lackey, to pick a couple of pitchers who haven’t had sensational seasons, not only have average Game Scores this year over 30, they have average Game Scores over 40—both of them. Nobody could stay in the league more than a couple of weeks if his Game Score average was 19, so a score of 190 doesn’t actually represent the level at which any pitcher is pitching.

For that matter, a score of 315 doesn’t actually represent the level at which any pitcher is pitching, either; I set it up that way because I think it is more fun to have a system in which a young pitcher has to work his way into the rankings by some sort of sustained success, rather than having a system in which he is “handed” a score of 400, but. . .it’s actually unrealistic to have scores of 315. (There would also be problems with starting the system at 400, and I’ll get to those in a moment.)

It’s not that no one would rank between zero and 300 if we allowed that; actually, lots and lots of pitchers would rank between zero and 300, but their ranking would not be a legitimate evaluation of their skill level for any of those pitchers, and this would cause huge and unsolvable problems for the system. The area zero to 300 would be populated by large numbers of

a) young pitchers who have not yet reached the level at which their scores represent their ability, and
b) old pitchers (out of the game) whose scores have not yet decayed to zero.

It would be impossible to get the rankings “right” in that zero to 300 range, because you’d be dealing with numbers that didn’t really represent any pitchers’ ability. Among the problems this would cause:

1) Kevin Millwood would rank higher than Stephen Strasburg, because Millwood’s scores from his past success would not yet have decayed to zero, while Strasburg would still be working to get his score up to 100.

2) Clayton Kershaw would not yet have reached the top 25 or, probably, the top 50. Because the system (starting at zero) would force a young pitcher to march across the “desert” from zero to 300 to enter the system, it would take a long, long time for young pitchers to reach the level at which their number represents their real skill level.
If you started at zero you would TEND to have a system in which fourth-year starters ranked ahead of third-year starters, third-year starters ranked ahead of second-year starters, etc.—not absolutely without exception, but in general. It would become a seniority contest, rather than an ability contest, because pitchers would take years and years to reach the point at which their number represented their real ability.

3) The “desert” from zero to 300 would be thickly populated by pitchers who had not yet disappeared from the system because their scores had not yet decayed to zero, but it would just be totally impossible to get these pitchers arranged in the right order because the numbers wouldn’t be meaningful.

We actually have all of these problems in the current system as I set it up, starting at 300. Suppose that I was able to sell this system to USA Today or somebody and the rankings actually appeared regularly, we would take criticism from people who would say “Look at this.. . . Rodrigo Lopez is here at 382, which ranks 117th, whereas Stephen Strasburg is at 318, which ranks 154th. That’s asinine; there is no way in the world Rodrigo Lopez is better than Stephen Strasburg. No manager who could choose Rodrigo Lopez or Stephen Strasburg to start a game tomorrow would choose Rodrigo Lopez.”
People would say that, and it’s absolutely right; it IS true that the system would have unrealistic relative values for those two pitchers. It’s a consequence of starting the system at 300.00, when the system REALLY should start about 425.00. But I’ll live with that, because
a) I think it is more fun to start the pitcher out at 300 and watch him work his way into position to compete, rather than giving him a competitive platform right away, and
b) We’re not really that focused on the down-list rankings. We’re MAINLY concerned, frankly, with making sure we’ve got the top 10, top 15 guys in a reasonable order; whether the guy who ranks 117th is better than the guy who ranks 154th isn’t so much a priority. And those rankings will change very quickly anyway. If Strasburg stays healthy and pitches well, he’ll rocket past Rodrigo Lopez in a month.

Plus, there are problems with starting people out at 425 (even though 42.5 is about the lowest Game Score that you can average and stay in the rotation.) If you make the “floor” to the system 425.00, then half the pitchers in baseball (or more than half. ..probably more than half) will fall back to 425.00 in the off-season, so then every season starts with 50 pitchers ranked and everybody else is at 425.00. This is unsatisfying; if you have 150 pitchers in rotation, you kind of want to have a list of 150 pitchers, rather than having 100 of them tied for 53rd place in the rankings at 425.00. So then we’d have to do something else with the off-season “decay” process, like terminating the adjustments on November 1, but that would just cause other problems.—ie, it would take forever for inactive pitchers who retired from a high platform (like Mike Mussina) to “decay” out of the system. It works better to live with the unrealistic scores in the range of 300 to 425.00—but it would totally unworkable to try to extend it out to zero. It’s one problem if Rodrigo Lopez ranks higher than Stephen Strasburg—which is wrong, but it will resolve itself in a matter of weeks. It’s a different issue if Kevin Millwood ranks higher than Stephen Strasburg, and Rodrigo Lopez still ranks higher than Stephen Strasburg a year from now. That’s what you’d have if you started the rankings out at 300.



2:16 PM Sep 21st
 
tangotiger
mskarpelos: the exponential decay is exactly what day-by-day Marcel uses.

The weight of each game is given: .999^daysAgo.

Now, you can try to accelerate that a little by using .998 instead.

You can limit it to the last 2 or 3 years if you want.

This is extremely easy to implement, and you don't have to worry about carrying over the previous data, into some rolling period.

You can also force in a Game Score of "30" or "40" for any of his missing starts, so that a player is penalized for not pitching.
1:23 PM Sep 21st
 
mskarpelos
I like the general approach, but I'm not sure I buy into game score as the appropriate metric or linear decay as the appropriate temporal weighting.

Wouldn't cumulative WAR (either the Fangraphs or the Baseball-Reference version) be superior to game scores? It seems to me that cumulative WAR is less subject to fluctuations than game scores.

Instead of a linear decay function why not a modified exponential decay? The linear decay function requires somewhat arbitrary constants whereas an exponential decay results in a smoother and more continuous decay. It's used successfully in such varied applications as modeling the half life of radioactive elements and block freshness in database caching algorithms. I'm sure we could adjust the exponent to properly model pitching performance.
6:29 PM Sep 20th
 
tangotiger
Chill: I proposed an enhancement to Game Score. Indeed, I have 4 different versions, each focusing on a separate aspect of pitching.

The readers (at Fangraphs, at my blog) then discussed their personal preference to weight those 4 versions.

Personally, I think this discussion can take place on Bill's site in the discussion area as well.


11:39 AM Sep 20th
 
glkanter
The immediacy of the internet is astounding and transparent.

If Bill had presented this 20 years ago, maybe a few people would have gone to the effort to write a letter with comments, complaints, praise, or suggestions.

Then it would have been a year later, if ever, that we saw any of that input.

Compare and contrast to what goes on today. And we wouldn't accept anything less than the instantaneous modes we have now.

For better or worse.
11:27 AM Sep 20th
 
chill
Two comments - the rankings Bill proposes work better for starting pitchers than they do for sports like tennis that make a big deal about rankings - in tennis, a player like Jelena Jankovic can end the year as #1 essentially because she played lots and lots of tournaments, and did pretty well. But even when she was #1, nobody would have picked her to beat, say, Serena Williams, a greater player who opted to play fewer matches that year. In baseball, there's no such thing as a lesser starting pitcher who starts 50% more games than a great pitcher.

Second comment - the weight game scores put on strikeouts is arbitrary, probably a bit high, and may perhaps give pitchers like Nolan Ryan a little leg up. But (adjusted) game scores still work, and I can't think of anything better.
10:50 AM Sep 20th
 
yorobert
great article, and i would love to see this list extended as far back as possible. this is baseball analysis that is fun.
7:02 AM Sep 20th
 
MarisFan61
To Sroney: Thanks for the comment! I think that explains it succinctly and fully.
12:06 AM Sep 20th
 
nettles9
This was fun to read. I think Bill is correct in just keeping this to starting pitchers since it would seem to be more upward and downward movement start by start.
5:41 PM Sep 19th
 
sroney
The percentage of decay that works is also affected by the baseline of 300, which takes off more points than would be taken off if the baseline were lower (as questioned by someone farther down the list).​
2:46 PM Sep 19th
 
tangotiger
#1: I like the premise.

#2: Using Game Score, or some variant of it, is the right way.

#3: The discussion point is on the weighting, the "decay" of past performance. Is current performance too heavily weighted? Should it be 2% or 4%, etc? That needs to be looked at more.

#4: How long would it take Strasburg to be #1, in the current era? Pedro at his peak averaged around a 70 Game Score. Let's say that Strasburg got a 70 for each and every start, and you start the clock back to April 1, 2010, and he never missed a start. Where would he be today?

#5: World's? It's MLB only, so, let's not say "world" while ignoring Asia completely.
1:57 PM Sep 19th
 
flyingfish
That is a great piece; thanks. I have long thought that Pedro was incredibly, dominatingly good and your rankings confirm that. I don't think I've ever seen anyone that good wearing a Boston uniform--and precious few wearing [i]anyone's[i] uniform--and I've been watching seriously since about 1973.​
12:33 PM Sep 19th
 
slideric
I think the good old days are now.....what a piece.......
shortstops would be interisting....thanks​
11:09 AM Sep 19th
 
joedimino
Retrosheet has boxscore information all the way back to 1918 - so you could start there.

Great concept. Would love to see the current rankings, or year end rankings for the last few years.

Could probably use that data to run a regression to estimate rankings for seasons where boxscore data isn't available.
9:29 AM Sep 19th
 
taosjohn
Kind of answers my question/concern about the 97 NL Cy Young...
9:19 AM Sep 19th
 
CharlesSaeger
Someone could do this going back to 1918, where Retrosheet has box scores. Actually, you could start with 1871, but you'll lose the scores after a few years so it's kind of silly.
8:59 AM Sep 19th
 
cderosa
I frickin' love this thing. It would be great to have it along with the Tired Closers list as something the site kept us posted on!
8:05 AM Sep 19th
 
greggborgeson
Great concept. Something easily accessible for the casual fan. I can imagine that this will add some late-season fun for fans of non-contenders as they watch a favorite talented young pitcher climb the charts. Let's hope the media pick this up quickly.
7:50 AM Sep 19th
 
MarisFan61
Please pardon if you explain this somewhere and I just missed it, or if it's somehow implicit except to dummies, but.....
Why do you choose "300" as the baseline? Why not zero?
I can imagine that maybe it's not zero because you don't want to ever bother with negative numbers in any part of the calculation, but it seems this could be avoided with a baseline of 100, which would feel like a more natural baseline, so why not that?

I wondered if maybe you did it so that the resulting numbers would correlate with some known number in a desirable intuitive way, sort of like how a starting pitcher's Win Shares are close to actual wins, but....it doesn't seem to be anything like that either.
What's the deal?
2:28 AM Sep 19th
 
rtallia
You're going to have to do this at least back to 1900 now, Bill--you know that, right? Great piece!
1:46 AM Sep 19th
 
glkanter
Man, that was fun! And informative! Took me back to the old days of the Abstracts!
12:40 AM Sep 19th
 
 
©2017 Be Jolly, Inc. All Rights Reserved.|Web site design and development by Americaneagle.com|Terms & Conditions|Privacy Policy