Following up on the little bit of work I did along this line a few weeks ago, I decided to do a more detailed and careful look at levels of quality games.  There are similarities to what I was doing last month and some differences. 

            About three years ago I did some research, as a part of which I derived an "expected Game Score" for every game.  The Game Score was 50, but modified by

a)     The league; if it is a pitcher’s league the expected Game Score is higher; if it is 1930 or 1999 the expected Game Score is lower,

b)     The park; if it is a hitter’s park the expected Game Score is lower, if it is a pitcher’s park the expected Game Score is higher, and

c)      The quality of the team the pitcher is facing.  If he is facing a weak-hitting team the expected Game Score is higher; if he is facing a strong offense the expected Game Score is lower.

I actually do not remember specifically how I derived those expected Game Scores, but I do remember that I double-checked or back-checked them thoroughly; in other words, checking to see that if the expected Game Score in a given circumstance was 47, the average of actual games played in those circumstances was in fact 47, so I have a lot of confidence that that part of the system works, although I no longer remember exactly how it was done.  Old research is like a highway that you drive over; you don’t always need to know how the highway was built. 

Anyway, I saved those numbers when that research was done.   The highest expected Game Score ever is 63.48, which was the expected Game Score when facing the 1963 Mets in Dodger Stadium—low run season, pitcher’s park, weak offense (Mets).  The lowest expected Game Score ever is 35.22, which was the expected Game Score when facing the 1936 New York Yankees in Shibe Park in Philadelphia.   High run season (the American League in 1936 is almost the same as the National League in 1930), formidable offense, and a hitter’s park.   Ivy Andrews faced the Yankees in Shibe Park on July 22, 1936, giving up 10 hits and 5 runs, all earned, 5 walks, no strikeouts, which under those circumstances is actually a good game.  Beat them 6-5, complete game. 

The average expected Game Score is 50.33, and the standard deviation is 3.38.  The average actual Game Score is 50.17, and the standard deviation is 17.69. 

In the little research I did last month, games were classified as Excellent Games based on the Game Score and the league; if the Game Score was in the top 20% of all Games Scores in that league, then it was counted.  In that study I used the expected Game Scores as a tie breaker; in other words, if 40 pitchers all had Game Scores of 66 but counting all of them would put us over 20% for the league, I counted the ones with the lowest expected Game Scores, since those pitchers posted a 66 in more difficult situations.  

For this study, I’m doing several things differently, but let me describe the study first.   Suppose that we measure six different levels of games for starting pitchers:

1)     SUPER GAMES are games in which the pitcher’s Game Score is at least 35 points higher than expected for the circumstances.  Almost precisely 1% of games are Super Games, as it turns out, although I didn’t plan that, but it’s 1.003%. 

2)     DOMINANT GAMES are games in which the pitcher’s Game Score is at least 25 points higher than expected for the circumstances.  About 7% of pitcher’s starts are Dominant Starts.

3)     EXCELLENT STARTS are games in which the pitcher’s Game Score is at least 15 points higher than expected for the circumstances.  Just over 20% of starts are considered excellent, just short of 21% but closer to 21 than to 20.

4)     GOOD STARTS are games in which the pitcher’s Game Score is at least 5 points higher than expected.  40% of starts are considered Good Starts, and we will consider these to be comparable to "Wins". 

5)     BAD STARTS are games in which the pitcher’s Game Score is at least 5 points WORSE than expected.  39% of starts by pitchers are considered Bad Starts.  Many pitchers have Good Start-Bad Start records which are quite similar to their won-lost records.

6)     TERRIBLE STARTS are games in which the pitcher’s Game Score is 25 points worse than expected. 

And reminding you about the conditions of the data file:  it includes games from 1921 to 2018 and is almost 100% complete since 1965, but missing more and more games as we go backward in time.

The most important ways in which this study improves over the work I did a couple of weeks ago are (1) that I am using an array of levels of games, rather than just looking at excellent starts, and (2) that I compared EVERY game to the expected game score for that situation, rather than comparing games to 50, and only using the expected Game Score as a tie-breaker.  Also, I’m not paying any attention to the 20% league standard that was the basis of the other study, although the "excellent game" standard is basically the same as the top 20% of games. 

I switched to Game Score-vs-Expected Game Score calculation because. . .well, because it is obviously better, since it adjusts for park effects and adjusts for the quality of competition you are facing and adjusts for the league, but more specifically I did that because Felix Hernandez looked SO great in the other study that I was wondering whether it was a park illusion or whether it would stand up to more careful inspection. 

I also made one other tiny little tweak to the process.  Some people will always say that the Game Score system overvalues strikeouts.  I don’t agree that it does.  People say that because old-line analysis, before Voros McCracken, wanted to treat a strikeout like any other out.  A pitcher strikes a batter out, or he gets a ground ball to short.  What difference does it make?  An out is and out.

An out is an out, yes, but the pitcher’s contribution to the two outs is not equal.  The pitcher who gets a ground ball to short has allowed contact by the batter which, in that case, resulted in an out, but which, in the larger picture, was at risk of becoming a hit, since the pitcher does not control the outcomes of balls in play against him.  With a strikeout, the pitcher has prevented the possibility of a hit by striking the batter out.  It’s not the same.

That’s the way I see it, but the "strikeout bias" argument is a reasonable argument, and it could be true to some as-yet-unmeasured extent.  I decided to accommodate that viewpoint to a certain degree, by subtracting .1 points from the Game Score for each strikeout.  Since starting pitchers within the data averaged 3.8 strikeouts per game, I then had to add 0.38 points per game back into the calculation, so that the scores would still center at 50.  So the formula for pitcher’s advantage, which I will refer to as Game Score minus expected Game Score, is actually Game Score + 3.8 – SO/10 - Expected Game Score. 

So given that background research, there are like a million things you COULD do with it, and it is kind of random where you start, so I’ll start here:

1.      Strings of consecutive good starts.

Pedro Martinez had a string of 30 consecutive good starts, beginning on August 3, 1999, and extending through August 8, 2000.  His start on August 14 against Tampa Bay was his only start in 2000 that would not be considered "good". 

Johan Santana had the second-longest streak in my data, 23 straight good starts in 2004.  Mike Scott had 20 straight in 1986 and Gaylord Perry had 20 straight good starts ending at the All Star break in 1974.   By the fourth of July, 1974, Gaylord had pitched 15 complete games and was 15-1 with a 1.31 ERA.  His streak ended in his first start after the All Star break, when he pitched into the 11^th inning against the Orioles, but gave up a 3-run homer in the 11^th inning. 

Clayton Kershaw had a streak of 19 consecutive good starts in 2014. 

2.      Strings of consecutive excellent starts. 

Pedro Martinez, as part of the same streak discussed before, had a stretch of 19 consecutive excellent outings.  That stretch ended on June 14, 2000, against the Yankees, when he pitched 6 innings, giving up 6 hits and only 1 run, striking out 7 and walking 2.  That’s a good start, but he only cleared the expectations line by 13.5 points in that one.  To be considered excellent, it has to clear by 15 points. 

Pedro has a 7-game lead on the field in terms of consecutive excellent starts.   Several pitchers have had streaks of 12 consecutive excellent starts.   Bob Feller had 12 straight in 1940, just after the trigger of the Cleveland Crybabies incident.  Feller had two straight not-especially good starts on a road trip to Boston and New York (June 7 and June 11, 1940), although he was still the best pitcher in the league at that time, but manager Ozzie Vitt loudly chewed him out in front of the team for giving up 5 runs to Boston (June 11).  When the team got back to Cleveland, Feller and a group of his teammates met with the Cleveland owner and demanded that Vitt be fired.  Vitt was not fired; it became a media circus, but Feller then ripped off a string of 12 consecutive excellent starts—the longest stretch in my data, other than Pedro in 1999-2000.  He completed all 12 starts, one of them in 12.1 innings, and posted a 1.46 ERA in the stretch. 

The others with 12 straight excellent starts:  Randy Johnson, 2000, Johan Santana, 2004, and Gaylord Perry in 1975 (a completely separate stretch from the one discussed above, in 1974.) 

3.      Strings of Consecutive Dominant Games

Well, here we learn something interesting that I didn’t know.  The longest streak of consecutive dominant games is 6, by Randy Johnson in 1999, Orel Hershiser in 1988, Javier Vazquez in 2001, and Johnny Vander Meer in 1938. 

I did not realize that Vander Meer’s two consecutive no hitters were part of a longer stretch of dominance.   The streak really began May 20, 1938, when Vander Meer pitched a 5-hit shutout with 9 strikeouts against the Giants, National League champions 1936-37-38.  He didn’t have a great game in his next start, so those two are not part of the streak, but on May 27 he beat the Cardinals 2-1 in ten innings, striking out ten.  That started the streak of six consecutive dominant games, during which Vander Meer pitched 55 innings, giving up only 17 hits and 4 runs, and pitching two consecutive no hitters.  He was also in the middle of an eight-game winning streak. 

Javier Vazquez, pitching for Montreal at the time, had a run of six straight dominant games in August, giving up 3 earned runs in 49 innings. 

4.      Streaks of Consecutive Super Games

Five pitchers in my data have thrown three consecutive Super Games:  Teddy Higuera in 1987, David Cone and Randy Johnson, in 1994, Johnson again in 1997, and Roger Clemens in 1998.

5.      Consecutive Bad Starts

You could make 100 straight good starts and nobody would complain about that except Josh Donaldson, but if you make too many straight bad ones the streak tends to come to an end with no additional input from the pitcher.  The most consecutive bad games that anyone has been able to get away with is 13, by Brian Matusz, between June 12, 2011, and April 20, 2012.   Rick Langford also had a streak of 13 straight, beginning September 27, 1982, and lasting until July 23, 1985.   Langford did not start again in 1985, but ended the streak in his first start of 1986, when his Game Score was only 4.7 points below expectation!!  It has to be 5 below.  But then he got pasted again in his second start of 1986, so we could say it was really 15 in a row. 

6.      Consecutive TERRIBLE starts.

The record for consecutive terrible starts is four, by Jim Lonborg in 1978, Sean O’Sullivan in 2013, Paul Spoljaric in 1999, and Early Wynn in 1942.

7.      Most Super Games in a Season

Only two pitchers in my data have pitched 7 Super Games in a season:  Lefty Grove in 1936, and Pedro Martinez in 2000.  No one has pitched 6; just those two guys at 7, and a bunch of guys with 5.

Lefty Grove in 1936 was 36 years old, and had an uncharacteristically poor 17-12 won-lost record.  He did, however, lead the league in ERA, shutouts, and strikeout/walk ratio, all of this despite pitching in a hitter’s park.  His 7 Super Games were his six shutouts and a 4-hit complete game against the Yankees, in Fenway, in which he struck out a season-high 11. 

With regard to this particular measurement, Grove in 1936 and Pedro in 2000 have an "advantage", so to speak, in that they are working in an extremely high run environment, which pushes the expected Game Score down into the low 40s (against a good-hitting team), so that there is more operating room at the top of the scale.  Against an average team in an average league, the expected Game Score is 50, so you have to hit 85 to have that top-one-percent game.  Pedro and Grove could do it with 80, or maybe even 77.   It’s only an advantage, of course, if you pitch really well; if you don’ t pitch well it’ll mess up your ERA.  But there is a mathematical edge for them in there somewhere.  (This doesn’t really apply to any lower standard, such as dominant games.  It applies in theory, but it is only actually meaningful in the most extreme cases.)

8.      Most Dominant Games in a Season

These pitchers pitched the most Dominant Games in a season:

            Pedro Martinez, 2000       18

            Randy Johnson, 1999        17

            Bob Feller, 1941                 15

            Sandy Koufax, 1963           14

            Dwight Gooden, 1985       14

            Tom Seaver, 1971              14

9.      So, Pedro

I feel like I am being pushed gradually here toward the unavoidable conclusion that Pedro Martinez in 1999-2000 was the greatest pitcher of all time, in the sense of who would you most want to have on the mound in a must-win game.  Pedro did not last as long as a great pitcher as Spahn or Clemens or Nolan Ryan or Seaver or Lefty Grove, and he did not carry as heavy a load in any season as Koufax or Robin Roberts or Nolan Ryan or Steve Carlton.   But in terms of dominating while he was at his peak, I think he has to be the #1 man.  We are studying GAMES here, rather than seasons, but not single games; groups of games.  He has turned up at the top of the list several times here already, and he will continue to do so as this article goes on. 

10. Duh

A "Dominant" game is a game in which the Game Score is 25 points better than expected; a "Terrible" game is a game in which the Game Score is 25 points worse than expected.

In my data there are only three pitchers who made only one major league start, but it was a Dominant game.  But there are more than 100 pitchers who made only one major league start, and it was terrible (duh). 

The three pitchers who made only one start, but it was a Dominant Game, are Luis Aloma (June 17, 1951), Dave Jolly (July 17, 1954) and Darrin Winston (September 15, 1997).  

The most interesting of those is Luis Aloma, whose odd-looking career will pop up in all kinds of studies.  In his career he was 18-3, which I think may be the best career winning percentage ever for a pitcher with 20 or more decisions (he was 7-2, 6-0, 3-1 and 2-0).  In 1951 he was 6-0 with a 1.82 ERA, but also he hit .350, so that pops up when you do a search for guys with fantastic cards for a table-top league.  He never committed an error in his career, so he has a career fielding percentage of 1.000. 

He made his only major league start in the second game of a double-header, the fifth game of a five-game series in which the White Sox (Aloma’s team) had lost the first three games.   Aloma pitched a 5-hit shutout, facing only 30 batters.   He was in the majors for two years after that, but never started another major league game.

11. Most Excellent Starts in a Season

There are seven pitchers within my data who had 24 or more Excellent Starts in a season:

Sandy Koufax, 1963           26

Pedro Martinez, 2000       25

Sandy Koufax, 1965           24

Sandy Koufax, 1966           24

Vida Blue, 1971                   24

Steve Carlton, 1972           24

Dazzy Vance, 1924             24

12. Most Good Starts in a season:

Eleven pitchers within my data have recorded 30 or more Good Starts in a season:

Sandy Koufax, 1966           33 out of 41

Denny McLain, 1968          33 out of 41

Sandy Koufax, 1965           32 out of 41

Mike Scott, 1986                32 out of 37

Sandy Koufax, 1963           31 out of 40

Ferguson Jenkins, 1971    31 out of 39

Steve Carlton, 1980           31 out of 38

Bob Feller, 1940                 30 out of 37

Nolan Ryan, 1973               30 out of 39

Mickey Lolich, 1971           30 out of 45

Don Drysdale, 1964           30 out of 40

Several of those are interesting seasons.  In 1971 both Cy Young Awards were disputed in the media, with many writers feeling that the NL Award should have gone to Tom Seaver, rather than Ferguson Jenkins, and some writers feeling that the AL Award should have gone to Mickey Lolich, rather than Vida Blue. 

You will notice that in the last few charts, we have outlined both sides of the case for both leagues.  Tom Seaver (NL) made 14 Dominant Starts in 1971, one of the highest totals ever, but Ferguson Jenkins (this made 31 good starts out of 39.  Vida Blue (AL) made 24 excellent starts, third-highest total in my data, but Mickey Lolich made 30 good starts.   I’m also happy to show a little support for Denny McLain’s 31-win season.   He didn’t dominate like Koufax or Carlton or Gibson, but he did have a consistency of excellence, making 33 Good Starts out of 41.  His Good Start/Bad Start ratio (33-6) is almost the same as his won-lost record. 

13. So, Koufax?

It is pretty common now to hear people under the age of 60 say that Koufax was terribly overrated.   The people who say this are speaking out of ignorance.  They don’t really understand who Koufax was or what he did. 

The Koufax-baiters will say (a) that Koufax was great for only a few years, which is true, and (b) that Koufax’ numbers were helped by pitching in a great park for a pitcher in a series of low-run seasons, which is true. 

But this lens that I am looking through now, this method takes the park effects and the league-offense-level effects out of the question by adjusting every game for the expected performance in that set of circumstances—and yet, there is Koufax, still at the top of the list in all three of his great seasons. 

Yes, Koufax was great for a relatively short number of years, compared to some others—but what people are missing is that his level of dominance is greater.  Vida Blue was that great for one year, and Carlton was for one or two years, and Guidry maybe for one year (although he didn’t pitch as many games), and Dwight Gooden for one year and Randy Johnson and Pedro for a couple of years.  Warren Spahn was great, and Greg Maddux was great, and Gibson was great, and Tom Seaver was great—but Koufax, in his best years, was greater.  No pitcher in the last 100 years has had a string of dominant seasons to equal Koufax from 1963 to 1966.   

And that’s not the ONLY reason he is unique and special; there’s another element to it.  He could win games 1-0 and 2-1, reliably, win almost all of them, like nobody else ever has.  And he had more impact on pennant races than any other pitcher of the 20^th century, except possibly Carl Hubbell. 

If you go 20-10, that’s a great season.   If you go 25-5, that’s a different thing.  If you go 25-5 not because the team scores 5.7 runs a game for you, like Whitey Ford in ’61, but because you are just way better than everybody else and you beat them every goddamned time you step on the mound, that’s a different level of greatness.  20-10 is ten games over .500; that’s great.  25-5 is twenty games over .500.  The Dodgers won three pennants that they would not have won if Koufax had been just Don Drysdale great, or if he had been just Bob Gibson great.   They won them because he was greater than great.  Nobody else can say that, except arguably Carl Hubbell. 

14. The most BAD starts in a season.

With apologies to my longtime friend Larry Taylor, who is also a friend of the accused, Rick Mahler in 1986 made 39 starts, of which 23 were bad ones. 

Kyle Lohse had 22 bad starts in 2004, and six pitchers in my data had 21 bad starts:  Jim Bibby (1974), Stan Bahsen (1972), Harry Byrd (1953), Denny McLain (1971), Steve Trout (1983), and Jose Lima (2005). 

15. Spahn and Sain in the Rain

The 1948 Boston Braves won the National League pennant led by two starting pitchers, Warren Spahn and Johnny Sain.  Their lack of depth in the starting rotation made famous the phrase "Spahn and Sain and pray for rain", although here is an odd fact for you.  Spahn and Sain had a combined won-lost record that season of 39-27, a .591 winning percentage.  The rest of the team had a won-lost record of 52-35, a winning percentage of .598.  So Spahn and Sain in 1948 actually had a worse record than the rest of the team.

Anyway, my point was that in order to set a record for the number of terrible games in a season, a pitcher has to begin the season carrying great confidence from his manager.   Johnny Sain in 1949, the next season, pitched TEN really terrible games, the most of any pitcher in my data.  

Actually, we could say it is 11.  Sain made his last start of the 1949 season on Thursday, September 29, 1949, against the Brooklyn Dodgers.  The Dodgers were one game out of first place with four games left to play, and a double-header to be played that day.  The Braves started Spahn and Sain, and it was, in fact, raining, raining pretty hard by the time the second game got underway.  The Dodgers pounded Spahn in the first game, beating him 9 to 2.  Sain started the second game.  He gave up five runs in the first, getting one man out.  The one guy he got out was Duke Snider, who popped out to third base while attempting to bunt.

Anyway, that’s a pretty terrible—five runs in one-third of an inning—but it is not QUITE terrible enough to make the list.  His Game Score was 22, but, because the Dodgers were a heavy-hitting team, the expected Game Score was only 46.4.  He was 24.4 points below expectation.  It has to be 25 to be counted as terrible.  It’s silly, I know, but you have to have some sort of rules. 

Anyway, we could really say that it was 11 terrible starts for Sain that year—I would—but I should note that actually pitched well in his other 25 starts, with an ERA a hair over 3.00. 

Other than Johnny (in) Sain, the most terrible starts by a pitcher in a season in my data was 9, by Eric Milton in 2005 and James Shields in 2016.   Many pitchers with eight.

Super Games in a Career

            Randy Johnson had 46 Super Games in his career, easily the most of any pitcher in my data.  He is followed by Nolan Ryan, 39, Roger Clemens, 38, and Pedro Martinez, 37.   No one else had more than 26.

            Super Games per start, Pedro is way ahead of everybody else.  It’s about one Super Start for every 11 starts, or about 9 times the normal frequency.

Dominant Games in a Career

            Randy Johnson, 147

            Roger Clemens, 141

            Nolan Ryan, 132

            Tom Seaver, 112

            Bert Blyleven, 108

            Pedro Martinez, 106

            No one else over 100. 

Doesn’t Randy Johnson also have a case to be considered the most dominant pitcher of the last 100 years?

            Sure.  And you know what’s odd about that?   Pedro and Randy.  It’s the same era, in that they both were at their peak at about the same time, and the Expos had them both, and let them both get away. 

Excellent Games in a Career (excellent being 15 points better than Expected):

1.     Roger Clemens, 313

2.     Nolan Ryan, 296

3.     The Big Unit, 273

4.     Tom Seaver, 242

5.     Steve Carlton, 237

Good Starts in a Career:

1.     Nolan Ryan, 466

2.     Roger Clemens, 458

3.     Greg Maddux, 415

4.     Steve Carlton, 390

5.     Randy Johnson, 389

Bad Starts in a Career:

1.     Tommy John, 255

2.     Jim Kaat, 245

3.     Jerry Reuss, 231

4.     Jamie Moyer, 222

5.     Don Sutton, 219

Terrible Starts in a Career:

1.      Don Sutton, 56

This is like the record for strikeouts.  The record for strikeouts in a career has been held by Babe Ruth, Mickey Mantle and Reggie Jackson—the point being that you don’t get to do that much bad stuff in a career until you are doing a lot of Good Stuff along the way.  Cy Young holds the career record for losses. 

Don Sutton in his career had a Good Game/Bad Game percentage of .635 (381-219).  That’s really good.  But he had a long, long career—I think he holds or used to hold the career record for starts—so some of them weren’t pretty.

1.      Don Sutton, 56

2.     Mike Moore, 54

3.     Tommy John, 53

4t. Early Wynn, 51 (Significant Data missing)

4t.  Rick Wise, 51

6. Gaylord Perry, 48

Good Game/Bad Game Percentages

            I mentioned before that many pitchers have Good Game/Bad Game percentages in a season that resemble their Won-Lost records.  That’s true, and then again it isn’t.  As the number of decisions gets larger, the Wins and Losses tend to separate from the Good Game/Bad Game versions of them, and also, as you get further from .500, the records tend to move apart.  So if you are talking about a 30-game sample from a kind-of-average pitcher, the records will usually be recognizably similar, if not identical, but if you are talking about a 400-game sample from a great pitcher, they won’t be.   

            There’s actually a really interesting phenomenon here which I have noted before.  A pitcher’s career won-lost record tends to be the half-way point between .500 and his percentage of good games.  In other words, if a pitcher pitched well in ALL of his starts, he would tend to have a winning percentage about .750.  If he pitches well in 70% of his starts, he will have a winning percentage about .600—depending on his offensive support and on other factors, of course, but in general.  (.500 + .700) / 2 = .600.    It’s interesting, because it suggests that a pitcher’s won-lost record is about 50% dependent on how well he pitches, and about 50% dependent on other factors. 

            Anyway, among pitchers with at least 200 Good Games in their careers, there are 12 who have Good Game Percentages (within my data) of at least .700: 

Greg Maddux was 415-181, a .696 percentage.  Bret Saberhagen just misses; he is 199-88, or .693.  Regarding shorter careers. . .with 100 Good Games but not 200, the leading percentage is Johan Santana, 177-51 for .776.  He is followed by Brandon Webb (120-44, .732), JR Richard (120-49, .720), Kerry Wood (101-41, .711), Lefty Gomez (missing data, 161-67, .706), and Sandy Koufax (187-79, .703). 

            And, among pitchers with yet shorter careers, 50 or more Good Games, we have three with .700 percentages:  Jose Fernandez (51-12, 810), Josh Johnson (87-32, .731) and Mark Prior (58-22, .725). 

The Greatest Pitcher of All Time?

Another thing you can do with this process is to find the sum of the pitcher’s Game Scores minus expected Game Scores for his career.  The greatest pitcher of all time would presumably be the pitcher with the highest total.  That would be Roger Clemens, who was in fact the greatest starting pitcher of all time.  Clemens beats everybody else by more than 10%, everybody except Randy Johnson by more than 20%.  If I had complete career data for Cy Young, Walter Johnson, Pete Alexander and Christy Mathewson, I assume that some of them would probably beat the Rocket. 

1.      Roger Clemens             +7,177 points vs. expectation

2.     Randy Johnson             +6,433

3.     Nolan Ryan                    +5,787

4.     Pedro Martinez             +5,092

5.     Tom Seaver                    +5,074

6.     Greg Maddux                +4,257

7.     Lefty Grove                    +4,108

8.     Curt Schilling                  +3,988

9.     Bert Blyleven                 +3,956

10.Bob Gibson                    +3,773

The Worst Pitcher of All Time?

            Of course, nobody sticks around long enough to be 7,177 points WORSE than expected in his starts, but Mike Pelfrey did the worst he could.  Note the odd similarity between Pelfrey’s number and Clemens—7,177 and 1,1171.   Pelfrey, a 6-7 right-hander from Wichita State, had ERAs over 5.00 seven times in a twelve-year career:

1.      Mike Pelfrey                 -1,171 points vs. expectation

2.     Jack Fisher                      -1,026

3.     Kevin Correia                   -964

4.     Jeff Suppan                      -917

5.     Mike LaCoss                     -891

6.     Jamey Wright                  -885

7.      Kyle Kendrick                  -877

8.     Andy Hawkins                  -844

9.      Jason Marquis                 -811

10.  Jaime Navarro                -805

The Greatest Seasons of All Time, by the same method:

1.      Pedro Martinez, 2000            +779 points

       101 points better than any other season

2.     Randy Johnson, 1999              +678

3.     Pedro Martinez, 1999             +662

4.     Sandy Koufax, 1965                 +658

5.     Roger Clemens, 1997              +644

6.     Dazzy Vance, 1924                   +635

    (missing a few games for Vance)

            7. Pedro Martinez, 1997               +630

            8.  Randy Johnson, 2001              +625

            9.  Dwight Gooden, 1985             +599

            10.  Steve Carlton, 1972               +599

            11.  Sandy Koufax, 1966               +598

            12. Randy Johnson, 2002             +597

            13.  Vida Blue, 1971                       +591

            14.  Bob Gibson, 1968                   +587

            15.  Mike Scott, 1986                    +578

            16.  Ron Guidry, 1978                   +574

            17.  Randy Johnson, 2000            +573

            18.  Bob Feller, 1940                     +572

                        (Missing a few games for Feller)

            19.  Tom Seaver, 1971                  +569

            20.  Randy Johnson, 1995            +565

            21.  Sandy Koufax, 1962               +562

            22.  Randy Johnson, 2004            +553

            23.  Steve Carlton, 1980               +549

            24.  Nolan Ryan, 1973                   +546

            25.  Bob Feller, 1939                     +543

                        (Missing a few games for Feller)

This is NOT a Cy Young Discussion; no, it isn’t.

            There are a million methods to re-visit Cy Young Awards, and I don’t know that the world would profit from adding another one.  We can, however, look quickly at a year-by-year list of the best starting pitchers in each league each year. 

            Reminding you again that we are missing spotty data up until the mid-1950s and occasionally missing a start after that, there are 42 pitchers who have ranked as the #1 pitcher in their league at least twice.   I will run down the list of those, and if a pitcher is not mentioned here, it is because he was not the best pitcher in the league, by this method, more than once.  

            Also, for convenience, I am going to say that the pitcher "was" the best pitcher in his league in that year, although of course what I really mean is merely that, by this method, he ranks as the best.

            Urban (Red) Faber was the best pitcher in the American League in 1921 and 1922.  Most pitchers who were the best pitcher in the league in two years and only two did so in consecutive seasons, although there are some exceptions to that rule. 

            Dolf Luque was the best pitcher in the National League in 1923 and 1925.   In 1925 he was just 16-18 for a team that was 80-73. 

            Dazzy Vance was the best pitcher in the National League in 1924, 1928 and 1930.

            Lefty Grove was the best pitcher in the American League eight times—1926, 1928, 1929, 1930, 1931, 1932, 1935 and 1936.  Only one other pitcher has had as many as eight seasons as #1.   Until about 1990, when the great pitchers of the 1990s emerged (Maddux, Clemens, Randy Johnson and Pedro Martinez). . but until then, I thought Grove was the greatest pitcher in baseball history.  Even later in his career, when Grove could no longer carry the workload to be the most valuable pitcher in the league, he would still lead the league in ERA and win almost every decision. 

            Carl Hubbell, AKA King Carl, ranks as the best pitcher in the National League in 1932 and 1936, but oddly enough does NOT ranks #1 in 1933, when he was the National League MVP (as he was in 1936).  This system goes with the Giants #2 starter, Hal Schumacher.  I assume that this is just because of some missing data. 

            Lefty Gomez was the best pitcher in the American League in 1934 and 1937, pitching about 280 innings with a 2.33 ERA both times. 

            Bucky Walters was the best pitcher in the National League in 1939, 1940 and 1944.  A converted infielder, he also hit .325 in 1939. 

            Bob Feller was the best pitcher in the American League in 1939, 1940, 1941 and 1947. 

            Mort Cooper was the best pitcher in the National League in 1942 and 1943.

            Hal Newhouser was the best pitcher in the American League in 1945 and 1946.   I may be missing some data from 1944; I’m not sure. 

            Ewell Blackwell was the best pitcher in the National League in 1947 and 1950.

            Early Wynn was the best pitcher in the American League in 1950, 1951 and 1954. 

            Warren Spahn ranks as the best pitcher in the National League in 1951 and 1956. 

            Robin Roberts was the best pitcher in the National League in 1952, 1953, 1954 and 1955.

            Billy Pierce was the best pitcher in the American League in 1953 and 1955.  His 20-win seasons were later, 1956 and 1957. 

            Don Drysdale was the best pitcher in the National League in 1957, 1959, 1960 and 1964, but, like Pierce, does not rank #1 in either of his 20-win seasons, 1962 and 1965. 

            Jim Bunning was the best pitcher in the American League in 1957 and 1960, and in the National League in 1967.  The 1960 Bunning season has always fascinated me because Bunning was 11-14, and was not perceived as anywhere NEAR the best in the league.  But it was a great year; he just didn’t get the wins.           He was also the first pitcher to rank #1 in both leagues. 

Whitey Ford, by this method, was the best pitcher in the American League only once, in 1958, when he was just 14-7 but threw seven shutouts and had a 2.01 ERA.   The 1961 season is an odd one in which you get a different answer to the question "Who was the best pitcher in this league" every time you study the question.   This method is going to say that it was Camilo Pascual. 

            Camilo Pascual was the best pitcher in the American League in 1959 and 1961. 

            Sandy Koufax was the best pitcher in the National League in 1961, 1963, 1965 and 1966. 

            Bob Gibson was the best pitcher in the National League in 1962, 1968 and 1969.  In 1962 Gibson was just 15-13 and nobody thought he was GREAT, but he pitched a lot of really good games in there. 

            Gary Peters was the best pitcher in the American League in 1963, 1966 and 1967.  Peters is probably the most under-appreciated pitcher on this list.  He was also a good hitter, hitting at least one home run every year from 1963 to 1971, and driving in 19 runs in 1963 and in 1971.  1964 was actually his best season, but Dean Chance was better that year.

            Sam McDowell was the best pitcher in the American League in 1965 and 1970. 

            Tom Seaver was the best pitcher in the National League in 1970, 1971, 1973, 1975, 1976 and 1977.

            Gaylord Perry was the best pitcher in the American League in 1972 and 1974.

            Steve Carlton was the best pitcher in the National League in 1972 and 1980.  Carlton won four Cy Young Awards, but does NOT rank by this method as the best starting pitcher in the league in 1977 (Seaver) or 1982 (Mario Soto).

            Nolan Ryan was the best pitcher in the American League in 1973 and 1977, and in the National League in 1987.    In 1987 he led the league in ERA and strikeouts, but finished 8-16. 

            Phil Niekro was the best pitcher in the National League in 1974 and 1978. 

            Ron Guidry was the best pitcher in the American League in 1978 and 1979. 

            Mario Soto was the best pitcher in the National League in 1982 and 1983. 

            Dave Stieb was the best pitcher in the American League in 1982, 1983, 1984 and 1985. 

            Dwight Gooden was the best pitcher in the National League in 1984 and 1985. 

            Roger Clemens was the best pitcher in his league an incredible 11 times, three more times than any other pitcher.  He ranks as the number one pitcher in his league in 1986, 1987, 1988, 1990, 1991, 1992, 1994, 1996, 1997, 1998 and 2005.  I believe, without checking, that all of those except 2005 were American League, and I believe that that includes a string of years that were perceived at the time as bad seasons for him.

            Greg Maddux was the best pitcher in the National League in 1992, 1993, 1994, 1995 and 1998. 

            Randy Johnson was the best pitcher in the American League in 1993 and 1995, and the best pitcher in the National League in 2000, 2001, 2002 and 2004.  In 1999 he ranks as the best pitcher in baseball but not as the best pitcher in either league, because he split the season between leagues.  But I’m going to count it for him anyway, which gives him 7 seasons as the best pitcher in his league, the most of any pitcher except Clemens and Lefty Grove. 

            Pedro Martinez was the best pitcher in the National League in 1997, and in the American League in 1999, 2000, 2002 and 2003. 

            Johan Santana was the best pitcher in the American League in 2004, 2005, 2006 and 2007. 

            Tim Lincecum was the best pitcher in the National League in 2008 and 2009. 

            Roy Halladay was the best pitcher in the American League in 2008 and in the National League in 2010.   He also won the Cy Young Award in the American League in 2003, but this system says that that one should have gone to Tim Hudson, who finished fourth in the voting.

            Felix Hernandez was the best pitcher in the American League in 2010 and 2014, winning the Cy Young Award in 2010 and finishing second in 2014. 

            Clayton Kershaw was the best pitcher in the National League in 2011, 2012, 2013, 2014 and 2015.   I believe that he is the only pitcher ever to rank as the #1 pitcher in his league for five consecutive seasons. 

            Justin Verlander was the best pitcher in the American League in 2011, 2012, and 2016, and in the National League in 2018.  I would assume that he was in 2019 as well, although I don’t have the numbers to support that. 

            Max Scherzer was the best pitcher in the American League in 2013, and in the National League in 2016, 2017 and 2018.  He is still pretty good. 

Worst seasons ever by a starting pitcher:

            Jose Lima with the Royals in 2005 was 5-16 with a 6.99 ERA.  The sum of his starts is 326 points worse than expectation, which is the lowest season total ever within my data.  These are the bottom ten:

Livan Hernandez is the only pitcher on the list with a winning record.  He pitched most of that season for the Twins, who would have won the division if they hadn’t had Livan Hernandez in the rotation most of the year.  They kept Hernandez in the rotation until early August, wound up losing the division by one game. 

Warren Spahn, aged 43, is the only Hall of Famer on the list. 

The Worst Starting Pitchers of All Time, by the ratio of Good Starts to Bad Starts

I think I used 200 starts as the cutoff to qualify for that list.  Of course, if you use a cutoff of 150 starts, you’ll probably find several pitchers with worse ratios than Trout.  That’s why I limited the list to five pitchers; it’s kind of arbitrary who you include. 

Pitchers whose Good Game/Bad Game Splits are far better than their Won-Lost records.

            Nolan Ryan completely dominates this category.  Ryan was 318-291 as a starting pitcher in his career (6-1 as a reliever).  But by this method, Ryan had 466 "good" starts in his career, 195 bad ones, and 112 which were in the middle, not really good or bad.  That means that he was shorted by 148 wins, and burdened unfairly with 96 losses.  That makes his won-lost record 122 games worse than his Good Game/Bad Game split.  

            No one else in history is remotely close to that.  Roger Clemens is second on the list, at -76.5, and Bert Blyleven third at -76.  There are 11 pitchers in history who are -50 or more (or +50, depending on which way you want to look at it.)   But none of them are great pitchers who had bad records because they were on bad teams; all of them were pitchers who had terrific career won-lost records, but even better Good Game/Bad Game splits, because, with a great pitcher, the team is never as great as the pitcher.  In fact, all of them had better career winning percentages than Nolan Ryan did.  So back to Ryan.

Nolan Ryan?

At the time that I started writing about baseball, mid-1970s, Nolan Ryan was in some sense the touchstone of all baseball arguments, not just involving sabermetrics and related issues, but in general, among people who had never heard of sabermetrics.  How good exactly was Nolan Ryan?  That was the number one thing that baseball fans argued about.  Yes, he pitches a lot of innings, his ERAs are good, his strikeouts are phenomenal, he pitches no-hitters, but his won-lost records are mediocre.  So how good is he, really?  In 1979 he was 16-14 with the Angels, left the Angels via free agency.  "No problem," said his General Manager, Buzzie Bavasi.  "We’ll just have to find two 8-7 pitchers." 

This is not JUST the old argument about the validity of won-lost records; it is a more nuanced argument in Ryan’s case.  Yes, Ryan did have phenomenal positives, and lots of them.  He not only threw harder than any other pitcher in that era, he threw harder than anybody does now.   And he did it 300 innings a year. There was—and is—nobody else like him. 

But Ryan also had phenomenal weaknesses in his game—and lots of them.  He walked more people than a 6^th Avenue Walk Light.  Somebody once ask me if the margin by which Ryan holds the career record for strikeouts was the widest percentage for any category.  Easy answer:  the margin by which Ryan holds the career record for Walks is FAR wider than the margin by which he holds the record for strikeouts.   It wasn’t just walks; he committed huge numbers of errors, I am certain more errors than any other pitcher in the last 100 years, leading the league regularly.  He led the league many times in Wild Pitches, and was close when he didn’t lead the league.  He led the league in Stolen Bases Allowed regularly, although at the time nobody knew that except my readers, because I was the only person who counted them.  His rate of getting hitters to ground into double plays was barely half of the league average, just a little over half (for his career.) 

So it is a fair question:  does he lose games because his team doesn’t score runs for him, or does he lose games because he walks people, throws wild pitches, commits frequent errors, never gets the double play when he needs it, and is the easiest pitcher in baseball history to run on? 

I’m trying to re-create the argument from the 1970s and early 1980s, and remember this:  at that time we did not have any methods to add up all the positives and negatives and find the balance that represents a player’s value.  There wasn’t any such method, really; I had the Value Approximation Method, but that had obvious problems, and was never prominent in the discussions. 

Decades pass; radar guns are invented, new arguments arise, new superstars rise and fall, methods to assign values to players are invented and become popular, and Nolan Ryan fades into the distant past. 

This particular method argues loudly that Ryan was one of the greatest pitchers ever to put on a uniform, but was disadvantaged in the won-lost record by crappy teammates.  The other side of the argument, however, still has a pathway forward.  This analysis is based on the Game Score method, and the Game Score method rewards strikeouts, but does not count Wild Pitches, Errors, or Stolen Bases Allowed.  It does mark a pitcher down for walks, true, but who is to say whether it marks a pitcher down ENOUGH for walks, or too little?  In this particular study I discounted the value of a strikeout from 1 point each to 0.9, but who is to say that that is ENOUGH of a discount?

I believe the conclusion that a strikeout is essentially of equal value to a walk is consistent with modern analysis, but you are free to argue to the point.  If you marked Ryan down one point for each error, one point for each stolen base allowed and one point for each Wild Pitch, that would move Ryan down the list, but Ryan is not the only pitcher who threw Wild Pitches and committed errors and allowed stolen bases, so it’s not clear how much ground he would lose compared to other pitchers. 

Pitchers whose Good Game/Bad Game records were far worse than their Won/Lost records

            The number one guy is Kirk Rueter, incongruously pronounced Reader, whose career won-lost record was 129-92, but whose Good Game/Bad Game Count is 117-138.   He would lose 29 games if we replaced his Won-Lost record with the alternative. 

            Rueter Reader is followed by the middle Grimsley, Ross Grimsley Jr., who would lose 24.5 games (122-97 to 104-128), then Bob Walk (-23), Tony Cloninger (-23),  Jerry Reuss (-22.5), Mike Torrez (-22.5) and Lew Burdette (-20). 

            Focusing on Seasons rather than careers,  Braden Looper in 2009 had a 14-7 won-lost record despite a Good Game/Bad Game split of 6-19, the largest discrepancy between those two in my data.  That’s a difference of ten games between his won-lost record and his Good Game/Bad Game split.  The interesting thing is that nobody bought it.  He was a free agent that winter, and did not receive any significant offer. 

            If he had done that ten years earlier, certainly if he had done that 20 years earlier, he would have received offers.   Storm Davis in 1989 went 19-7 despite a 4.36 ERA and a Good Game/Bad Game split of 12-15.  He was a free agent, and the Royals proudly snarfed him up with a multi-year contract, answering questions with "We’re not interested in ERAs; we want to win ball games."  But by 2009, everybody knew better.  Looper never threw another pitch in the major leagues. 

            Following Looper on the list:  Russ Meyer in 1953 (15-4 vs. 11-17) and Jim Merritt in 1969 (17-9 vs. 11-19).  Others of note:  Remy Kremer , 1930 (20-10 v. 11-16) and Bob Welch in 1990 (27-6 vs. 16-10). 

Greatest Starting Rotations in my Data

            The answers here are almost too obvious to discuss.  The top 10 rotations (in terms of Games Scores above Expected Game Scores) are:

1.      1998 Braves.   Maddux, Glavine, Smoltz, Millwood and Neagle.  +1195 points, which is 90 more than any other team.  Team won 106 games.

2.     1997 Braves.  Maddux, Glavine, Smoltz and Neagle.  +1104.

3.     2011 Phillies. Roy Halladay, Cliff Lee, Cole Hamels, Roy Oswalt, Kyle Kendrick and Vance Worley.  +1088.

4.     1980 Oakland A’s.  Mike Norris, Langford, Keough and McCatty.  +1085

5.     2016 Cubs.  Jon Lester, Jake Arrieta, Kyle Hendricks, John Lackey and Jason Hammel.  Had you forgotten already?

6.     2002 Diamondbacks.  Randy Johnson and Curt Schilling were the best two starting pitchers in baseball that year.  Rest of the rotation not so much.

7.     1993 Braves.  Maddux, Glavine, Smoltz and Steve Avery.  +1018

8.     1998 Yankees.  David Cone, David Wells, Andy Pettitte, Hideki Irabu, Orlando Hernandez and Ramiro Mendoza.

9.     1975 Dodgers.  Andy Messersmith, Don Sutton, Burt Hooton and Doug Rau.

10.2018 Houston Astros.   Justin Verlander, Dallas Keuchel, Gerrit Cole, Charlie Morton and Lance McCullers. 

Eight of those ten are in the last 30 years.  When I have drawn up similar lists in the past, years ago, the list would be led by the 1954 Indians, 1966 Dodgers and 1971 Orioles.   None of those teams show up on the list by this method; not saying that one method is right and the other wrong, but I will say that these are ten great pitching staffs no matter how you look at it.   If we stretch the list to the top 25, we get a good many older teams included—the 1930 Brooklyn Dodgers (or Robins), the 1931 Philadelphia A’s, 1944 Cardinals, 1948 Indians, 1956 Indians.  The 1954 Indians, 1966 Dodgers and 1971 Orioles still don’t appear.  I’ll re-evaluate. 

The Worst Starting Rotations in my Data

1,  1939 Philadelphia Athletics         -1102  in 152 games

2. 2013 Minnesota Twins                   -1136 in 162 games

3. 1939 St. Louis Browns                    -952 in 137 games

4. 1949 St. Louis Browns                    -731 in 108 games

5. 1996 Detroit Tigers                         -1047 in 162 games

6.  1954 Philadelphia A’s                    -988 in 153 games

7. 1998 Florida Marlins                       -1045 in 162 games

8. 1926 Phillies                       &nbs​p;              -731 in 114 games

9.  1957 Washington Senators         -973 in 152 games

10.  1937 St. Louis Browns                -895 in 142 games

So, Felix Hernandez?

Success in this metric correlates well but not exceptionally well with Hall of Fame acceptance.  There are so many very recent pitchers who stack up exceptionally well against history that we have to treat the values assigned to recent pitchers with some skepticism. 

That said, Felix Hernandez sort of looks like a Hall of Famer—not first in line for the position, not the greatest pitcher since 2000 or anything, but he looks good compared to historic standards.  Comparing his Game Scores to expected Game Scores in his starts, he would rank through 2018 as the 35^th greatest starting pitcher of all time. 

Of course, his 2019 performance (his last season, which was awful) would drag him down somewhat.  Un-elected pitchers ranked ahead of him start with Clemens and Schilling, and then the active pitchers:  Verlander, Kershaw, Greinke, Scherzer. 

But also clearly ahead of King Felix, by this method of measurement, are:

1)     David Cone

2)     Sabathia

3)     Santana

And, interestingly enough, Cole Hamels.  Thus, this method does NOT clearly conclude that Felix Hernandez should be next in line for the Hall of Fame.

But he still looks really good.  This method puts him ahead of Tiant, Bunning, Jack Morris, Lefty Gomez, Tom Glavine, Catfish Hunter, Red Ruffing, Ted Lyons, and presumably some others further down the list.  Bob Lemon and Early Wynn.

We have him with a Good Start/Bad Start ratio of 231-109.  Drysdale was 241-141.   Whitey Ford was 234-123.  So the method clearly does not say that Hernandez does NOT belong in the Hall of Fame, either.  The method says that we will need to look at the question from more different angles to get a real understanding of it. 

Thank you all for reading.