IV.   Morgan and Burkett

We are dealing here with the issue of whether some pitchers may have an ability to pitch especially well when they have an opportunity to win the game.  Let us contrast Mike Morgan with John Burkett.   Morgan made 411 starts in his career; Burkett made 423.    Morgan had a career ERA of 4.23; Burkett, of 4.31.   Morgan allowed 4.65 runs per nine innings, earned and un-earned; Burkett allowed 4.67.   Morgan pitched from 1978 to 2002: Burkett, from 1987 to 2003.   The two were teammates with the Rangers in 1999.

Yet while Morgan was 45 games under .500 in his career, Burkett—with essentially the same numbers, in the same era--was 30 games over .500.   Yes, offensive support did have a good deal to do with that.   Burkett did pitch for better teams.   Had Burkett had the same won-lost record as the rest of his team in every season, his career record would have been 159-143, whereas Morgan, as we said yesterday, would have been 162-165.

But Burkett also pitched better when he needed to pitch better to get a win.

John Burkett was shut out 20 times in his career, and his personal won-lost record in those games was 0-19; Morgan was shut out 28 times, and his personal won-lost record in those games was 0-27.    Morgan took a bigger hit, yes, but the two pitchers’ winning percentages, when they were shut out, were the same:

Of course, all pitchers have a .000 winning percentage when they are shut out, it being impossible to win those games.

Morgan also had more games in his career in which he was limited to one run, 42 to 33.   But you can win a game with one run, if you pitch well enough, and Burkett did this more times in his career than did Morgan:

Advantage, Burkett—a one-game advantage.

Mike Morgan had 65 starts in his career in which his team scored 2 runs; Burkett had 61.  Burkett won 13 of those games; Morgan, 3.

Advantage, Burkett—and this time it’s a big one.   Given just 2 runs to work with, Burkett rose to the occasion 13 times, Morgan only 3 times.   It looks like a ten-game advantage for Burkett, although it’s actually less than that.

Advantage, Morgan—a one- or two-game edge.   But working with four runs:

Advantage, Burkett—another three games or thereabout.  Working with five runs:

Advantage, Burkett.   Working with six runs:

Advantage, Burkett.   Morgan does have a one- or two-game advantage when working with seven runs:

But when working with eight or more, Burkett is back ahead:

V.  Method

Burkett clearly did a better job of rising to the occasion, over the course of his career, than did Morgan, and this clearly had value to his teams.   But how do we quantify that value?

Of course, I have examined this issue many times before over the course of the years, I think most recently in something published here about Bert Blyleven and Don Sutton.   But while I have studied this issue before many times, the two new methods that I have to offer here are, I am confident, the best methods I have ever developed to address this issue, this “ability to respond” by a starting pitcher. 

Here’s what I did.   What we’re trying to calculate here is the pitcher’s “effective runs allowed rate”, based on his ability to win the game with a given level of offensive support.

Of course, no pitcher has any ability to win the game when his team is shut out, therefore it’s not relevant data to what we’re trying to calculate, therefore we’ll set that data aside for right now.

At the other levels of run support, we’re going to work not with the pitcher’s individual won-lost record, but rather, with the won-lost record of his team.   I believe that this gives us a better working platform.   Suppose that two pitchers both have won-lost records of 16-4 when supported by 7 runs, but that one pitcher’s team is 17-5 in those games, while the other pitchers’ team is 17-11.   The bullpen has lost the game for his team six more times, but are those six extra losses relevant to establishing the performance level of the starting pitcher?

Yes, of course they are.   If you leave with the ballgame 7-6, there is a very good chance that your bullpen is going to blow the game for you.   If you leave with the ballgame 7-0, there is very little chance that the bullpen is going to give it away.   The team performance record, in my view, gives a more thorough look at what has really happened than the individual record.

OK, we’ll work with three pitchers here:  John Burkett, Mike Morgan and Whitey Ford.   When the team scored one run with this pitcher on the mound, John Burkett’s teams were 3-30, Mike Morgan’s teams were 3-39, and Whitey Ford’s teams, within the games documented by Retrosheet, were 10-28 with one tie. 

We know how many runs the pitcher’s team had to work with in those games, right?   It’s one run a game.   Burkett’s teams had 33 runs to work with, Morgan’s teams had 39, Ford’s teams had 38:

Now it’s Sabermetrics 101.   If Team A scored 33 runs in a set of games and their won-lost record was 3-30, how many runs did they probably allow?    Apply the Pythagorean Formula.   To get Burkett’s “effective runs allowed” we take the losses (30), divided by the wins (3), take the square root of that, multiply by 33:

Divide the runs allowed by the games, and you have the effective runs allowed rate:

What I am saying is not that Burkett allowed 3.16 runs 9 innings in these games, but that his teams won with the frequency expected if he allowed 3.16 runs per game.  

With 2 runs to work with, Burkett’s teams were 15-46, which is pretty good, but Ford’s teams were 19-22, which is really good.  Applying the same method, one can calculate the effective runs allowed rate for each pitcher, working with 2 runs:

ERAR standing for “Effective Runs Allowed Rate”; I despise acronyms, but sometimes you’ve got to do what you’ve got to do.  

And we can compare them when working with 3 runs:

At the level of three runs to work with, as mentioned before, Morgan was better than Burkett, although neither one of them was Whitey Ford.

VI. The Other Method

At this point we could rush ahead to our conclusion, but perhaps it is more fun to stop and admire the data a little bit, make some use of our other method, shake out the problems and challenges of the method, and then work back to our conclusion.

Of course, no pitcher won any games in which his team was shut out, but the pitcher who was shut out most often was Nolan Ryan, who was victimized by 66 shutouts in his career.   But these games are excluded from the studies of Effective Runs.

Working with one run, the number one pitcher in the data was Dean Chance, 1964 Cy Young Award winner.    Chance’s teams, in his career, were 17-41 when they scored just one run.   The overall winning percentage of teams that scored 1 run, in the Retrosheet data, was .101.   Chance’s teams were 11 games better than expectation in that situation.

These were the top ten pitchers in games with just one run to work with:

Chance individually was 16-35 in games with just one run to work with; his teams were 17-41.   These, on the other hand, were the pitchers with the worst won-lost records while working without a net:

Carl Morton was the National League rookie of the year in 1970; Steve Rogers could have won the same award for the same team three years later, although he didn’t.   Rogers certainly had a better career than Morton—but Morton’s teams were 9-22 when working with just one run, while Rogers’ teams were 1-46.    But working with two runs, Rogers was 22-37—a very good record—while Morton was 5-28.    Morton beats Rogers by ten games in games with one run; Rogers beats Morton by ten games in games with two runs.    The overall winning percentage of all teams with 2 runs to work with was .247, and these were the top ten pitchers in those games:

Tom Seaver pitched .500 ball when his teams scored just two runs (38-38), and his teams over-achieved in those games by twenty-plus games.   I don’t know that that’s more amazing than Mike Cuellar.   Cuellar was 27-21 in his career when his team scored two runs for him.  Now that’s impressive. 

On the other end of that scale was our friend Mike Morgan:

Let’s start tracking the aggregate performance.   Tom Seaver, based on his strong performance in games with two runs, is now in first place:

While Mike Morgan and Livan Hernandez are bringing up the rear:

I’m referencing a different method here, of course.   My basic, serious method for analyzing this data is the one I was working with in the last section, but to compile these charts, I’m doing something different.   Since the winning percentage of teams that scored two runs in a game was .247, I’m comparing each pitcher to a .247 winning percentage.   Tom Seaver, at two runs, was twenty games better than a .247 winning percentage.    There are problems with that method, which I will leave you to spot on your own, but it’s still useful, and as we will see it produces a list of the best pitchers which is pretty solidly identifiable as a list of the best pitchers.   

Three runs.   The number one pitcher, at being able to win the game with three runs, was Seaver’s American League contemporary and rival, Jim Palmer:

Palmer won almost two-thirds of his games, given three runs to work with, but Seaver is on the good list again, while Sandy Koufax, for a change, is not.   The overall winning percentage of starting pitchers working with three runs was .396.  But Pat Rapp, given three runs to work with, was 3-21:

Seaver is still in first place, Palmer moving up to third:

While Pat Rapp has now claimed the bottom of the chart.

Mike Morgan, with a positive performance at the three-run level (27-39), has escaped not merely the last spot, but the bottom 40.  Are you rooting for somebody here?   You should pick out somebody to root for; it makes the competition more fun.   And where’s Clemens, by the way?  Are he and Mindy in a back closet or something?

These are the overall winning percentages, for teams and starting pitchers, with each level of offensive support:

OK, four runs.   At the four-run level, the most effective pitcher at delivering a win for his team was Roger Clemens:

Clemens’ winning percentage, given four runs to work with, was just short of .750, both for him individually and for the team. 

In a way, these groups are like “years” of a player’s career.   We add up what the players did in 1991, in 1992, in 1993, etc.  Probably the same pitcher wasn’t the #1 guy each year; one year it was Clemens, another year it was Maddux, another year it was The Unit.   We add them together to form a picture of the whole.   Same here; some guys do great in one group of games, some guys in another.   It’s a competition.

In first place in the competition, right now, is Nolan Ryan:

Our leaders’ list is now a Hall of Famers’ competition.  The pretenders have gone home.    Dick Stigman some how managed to win only 3 of 21 games, given four runs to work with:

While Jeff Suppan Sanwiches has moved to the bottom of the summary competition:

Suppan was 4-35 with one run—average performance—but was 6-31 with two runs (-3 games), 14-33 with three runs (-4), and 25-36 with four runs (-7), quietly drifting to the very back of the list. 

Working with five runs, Tom Seaver returns to the top:

And this puts Seaver back in the driver’s seat in the overall competition:

Mark Clark, Aaron Cook and Brian Anderson managed to lose consistently with five runs of support:

Which puts Anderson in last place overall:

You will notice, however, that some of the totals are getting smaller.   At two, three and four runs, the leading pitchers were 20 games better than average.   At five runs, this number dropped to 12, and at six runs, it will drop to single digits:

Clemens had 70 games with six runs, still a very large number, but the “advantage/ disadvantage” numbers are shrinking rapidly because the overall winning percentage is approaching 1.000.    Clemens was 43-0 when he team scored six runs, but then, the overall winning percentage for starting pitchers working with six runs was .815.   Starting pitchers don’t lose a lot of games when the team scores six.   Although Mark Davis’ teams lost a lot:

Mark Davis was 5-17 as a starting pitcher in 1984, and there are ten losses here that didn’t go to him as a starter.  But he moved to the bullpen and won a Cy Young Award as a reliever.  With time running out, Tom Seaver solidifies his hold on the top spot:

While Jason Johnson has taken over the bottom:

Working with seven runs, the number one pitcher was Jim Kaat, whose teams were 44-2 at that level:

Ken Hill?   And look; Mike Cuellar is back.   The bigger tranches are more reliable, but the smaller ones are more fun.    Seaver, showing up on the leader’s list once again, pulls five games ahead in the overall competition:

Steve Trachsel’s teams lost 15 times when they scored 7 runs in a game:

Which puts Trachsel—and Sir Sidney—into the competition for the position as furthest below average:

OK, one more precinct to look at.   These were the leading pitchers when the team scored 8 runs, or more than 8:

While Pedro Astacio’s teams lost 13 times with 8 runs or more.  And there’s Jason Johnson again:

Which makes Johnson the lowest-ranking pitcher of the last sixty years, by this method:

Johnson, with a career record of 56-100, was 22 full games (44 half-games) under .500.   By this method, that somewhat understates how much damage he was really doing.   Tom Seaver heads up the Hall of Famers:

Seaver was 311-205 in his career, or 53 games over .500 (106 half-games).   This method puts him at +70.

This method, the method that I have been using over the last few pages, incorporates and is thus vulnerable to park and era illusions.   Tom Seaver looks a little bit better than he probably should because he pitched in a pitcher’s park in an era when ERAs were relatively low.   Clemens probably doesn’t look quite as good as he was, because he pitched in a high-run era.

Our other method, however, is not necessarily vulnerable to those problem.   In our other method, which we will return to tomorrow, we are figuring the “effective runs allowed rate” for pitchers by looking at their ability to win with one run, two runs, three runs, etc.   Of course, if a pitcher pitches in a hitter’s park and in a hitter’s era, his effective runs allowed rate will be different than if he had pitched in Dodger Stadium in 1965.

But in the comparison of John Burkett and Mike Morgan, for example, these issues are almost entirely irrelevant.   Burkett and Morgan allowed essentially the same number of runs per nine innings.   We then compare them on their ability to win with 1 run, 2 runs, 3 runs, etc.    Let us suppose that one of these pitchers pitched for much better teams, got much better offensive support, and pitched in a much better pitcher’s park.    So what?   It’s not (really) relevant (it’s marginally relevant for some minor issues.)   If pitcher A and pitcher B both allow 4.50 runs per game and are both working with 3 runs in a game, they should have the same ability to win, regardless of the park or the era in which they allowed these 4.50 runs and were supported by these 3 runs.    We’ll go back to that method tomorrow.