On July 1, 1956, Don Blasingame, leading off for the St. Louis Cardinals, had four hits in the game. Stan Musial also had four hits in the game for the Cardinals, while Wally Moon and catcher Hal Smith had three hits each. The other five spots in the batting order collected nine more hits, for a total of 23. One of Moon’s hits was a Grand Slam. Stan Musial also homered in the game, as did left fielder Rip Repulski.
In spite of all this hittin’, the Cardinals lost the game, 19 to 15. The Cardinal starting pitcher, Willard Schmidt, pitched 2 innings, giving up 6 hits and 5 runs. It was all downhill from there.
On June 27, 2007, the Kansas City Royals scored only one run in a game in which their starting pitcher, Jorge de la Rosa, gave up 10 hits in 6 innings of work. In spite of that, the Royals won the game, 1-0.
On May 18, 1955, the Cleveland Indians scored 19 runs in a game in which their starting pitcher, Herb Score, pitched a 3-hit shutout, striking out nine batters. Obviously, the Indians won the game.
On August 13, 2006, the Kansas City Royals were shut out in a game in which their starting pitcher, Luke Hudson, gave up 11 runs while retiring only one batter. Hudson’s game score, on what is normally a zero-to-one-hundred scale, was minus nine.
If your offense is shut out and your starting pitcher has a Game Score of -9, I would generalize, your team has little chance to win the game. Actually, we don’t need to generalize, because it doesn’t really happen very often; that’s actually the worst or second-worst combination of runs scored and starting pitcher performance in any game in the last 60 years.
I got into this by asking this question: What is a winning combination of Runs Scored and Starting Pitcher performance? Obviously if you score ten runs and your starting pitcher has a Game Score of 80, you’re likely to win the game. (More than "likely", as it turns out. There is no game on record in which a combination of ten runs and a starting pitcher Game Score of 80 did not result in a victory.)
Suppose that you take the number of runs scored in the game by the offense, multiply by 10, and add to that the starting pitcher Game Score. What is a Winning Combination of those two numbers?
90. Let us suppose that you predict:
a) That if the "Winning Combination Score" is 90 or more, the team will win, whereas
b) If the Winning Combination Score is 89 or less, the team will lose.
The Winning Combination Score being 10 * Runs Scored, plus the Game Score of the Starting Pitcher. If you make those that prediction, you will be correct 88% of the time.. . .actually 87.6%, so let’s say seven times in eight.
These numbers are (this formula is) essentially impervious to changes in Run Scoring Levels. 1968, as most of you know, was the deepest point of the run-scoring drought of the 1960s. But if, in 1968, you had predicted that every team that had a Winning Combination Score of 90 or higher would win and every team that did not would lose, you would have been right 90% of the time. ..actually, 89.97%. 2004 was more or less the heart of the steroid era, when many runs were scored. If, in 2004, you had predicted that every team with a Winning Combination of 90 or more would win and every team at 89 or below would lose, you would have been correct 87% of the time.
The formula "balances" across time, of course, because it has both a run-scoring and a run-prevention component. Let us say that your "90" is a combination of 3 runs scored by the offense (30 win points) and a starting pitcher Game Score of 60. Three runs scored by the offense are a "better" total in 1968 than they are in 2004, of course—but a Game Score of 60 is a better total in 2004 than in 1968. As one number gets "better" compared to the league, the other gets worse, so that the balance point stays about the same. Actually, the balance point goes to about 88 in 1968 and 91 in 2004, but… it doesn’t move very much.
If your Winning Combination Score is 120, what is your chance of winning the game? 91%. . .actually, .913. If your Winning Combination is 60, what is your chance of winning the game? It is essentially zero. In my data, the won-lost record of teams with a Winning Combination Score of 60 was 1 and 2304, which figures out to .000 even though there is a win in there.
This chart gives the winning percentage, in the data, at every Winning Combination Score up to 180:
The chart, as you can see, goes from near-zero to near-one very rapidly. At a Combination Score of 75, your winning percentage if .049. At 105, it’s .798. Around 90, the numbers move upward very rapidly. But when you’re not near 90, each point has very little impact.
The point of that is, most games are not in the "close call" range. About one-third of all team games have Winning Combination Scores of 106 or higher, about one-third are 74 and below, and about one-third are in the "competitive" range of 75 to 105.
I’m not trying to make this sound like something profound. It’s basically a truism, stated in numbers, the truism being that if you score more runs than you allow, you are likely to win. The games I started with at the top of the article were all extremely unusual games, from this standpoint. The Cardinals in the 1956 game had a Winning Combination Score of 176 (15 runs scored and a Game Score of 26), but failed to win—one of very few times, in the data, when a Combination Score that high did not yield a win. The Royals in the 2007 game had a Combination Score of just 62, which has a winning expectation of .004, but they were able to win that game, due to the unusual fact that de la Rosa gave up 10 hits in six innings, but no runs. (The "winning" games that score lower than 62 are almost all games in which the starting pitcher was injured and left the game in the first inning.)
The Indians in the 1955 game started by Herb Score had a Combination Score of 277 (19 runs and a Game Score of 87)—one of the better Combination Scores on record. And the Royals in the 2006 game had a Combination Score of negative nine, which narrowly misses being the worst combination on record. The one worse was an Oakland/New York game in 1998, in which Mike Oquist gave up 16 hits and 14 runs in a 14-1 loss, and I didn’t highlight that game here because I’ve written about the Oquist game before.
Picky, Picky, Picky
Of course some of you, who should be tortured before you are shot, are no doubt thinking that having a Combination Score of 115 based on 7 runs and a Game Score of 45 is different than having 4 runs of support and a Game Score of 75, and might yield a different win expectation for the team.
This is true, but not very true. If you think about it, if a system is 87.6% accurate at predicting whether a team will win or lose, there can’t be very much room to improve it by making simple adjustments. That’s why I didn’t experiment with 9 * runs and 8 * runs and 11 * runs. It’s just a back-of-the-envelope calculation. How much better than 88% do you think we’re going to get?
Just Didn’t Pitch Well
It occurred to me that, with the Winning Combination Score, one could very easily estimate to what extent each pitcher was responsible for the games he didn’t win. If it takes a combination of 90 to win a game and each run is 10 points, then it could be said that
a) the offense has entirely won the game if they score 9 runs or more, and
b) the starting pitcher has entirely won the game if he has a Game Score of 90 or higher.
Take, for example, the five losses of Justin Verlander in 2011, which was the year Verlander went 24-5. On April 11, Verlander pitched a six-hit complete game, losing to Texas 2-0. Obviously that’s not really his fault. His Game Score was 70. Verlander fell short of perfection by 20—two runs—whereas the offense fell short of perfection by 90, or nine runs. The offense was 82% responsible for that defeat (90 / 110), whereas Verlander was 18% responsible for it.
Verlander dropped his next start as well, losing to Oakland 6-2 on April 16. That one was more on Verlander’s shoulders than the Texas game, and that one is probably the reason Kate Upton didn’t like him. His Game Score was just 45, which is 45 short of guaranteeing a win; still, the offense fell short by seven runs, so that one is still 61% the responsibility of the offense (70 / 115), and 39% the responsibility of Verlander (45 / 115).
Verlander won one game and then lost again, dropping him to 2-3. He lost to Seattle 10 to 1, the bullpen giving up 7 of the 10 runs. Verlander had the oft-questioned Quality Start, 3 runs in 6 innings, a Game Score of 53. Again, the offense fell short of the Win Standard by 80 points (8 runs); Verlander fell short by 37 points. The offense is 68% responsible for that loss (80 of 117); Verlander is 32% responsible.
Verlander ripped off a string of wins then, lifting him to 11-3, finally losing to the Angels and Danny the Rabbit (Haren), 1-0 on July 5. This one was almost entirely on the offense, which did nothing, which is 90 points short of the Winning Standard. Verlander gave up 7 hits and one run in 7 and two-thirds innings, striking out 8, Game Score of 67. Verlander was 20% responsible for that defeat, the offense 80% responsible (90 / 113).
Verlander Just Didn’t Pitch Well, didn’t pitch up to the standard of the rest of the season, in his final loss of the year, July 15 against the White Sox. In that one he gave up five runs in six innings, Game Score of 44, losing the game 8-2. That one is 46 points on the shoulders of the ‘lander (90 – 44), and 70 points on the shoulder of the team (9 – 2, times 10). So even that loss is 60% the responsibility of the team, 40% Verlander. There actually aren’t any games in 2011 that Verlander Just Didn’t Pitch Well.
You are probably thinking, at this point, that the system gives most of the blame for defeats to the offense, which is not really true; Verlander isn’t a typical pitcher, and he actually pitched pretty well even in the games that he lost. My system does hold the offense responsible for 54% of games that starting pitchers don’t win and for 52% of their losses. I had a more complicated version worked out in which the most important runs scored were the first runs scored, and this gave most of the responsibility to the starting pitcher, but. . .it’s just a little messing-around toy, as opposed to a serious analytical tool. I liked the simple version that was easy to explain better, but the more complicated version did hold pitchers 60% responsible for the games they didn’t win.
I figured the "Didn’t Pitch Well Percentage" for every pitcher in my data. ..basically everybody in the last 60 years. I figured the Didn’t Pitch Well Percentage both for losses and for no-decisions, but weighted each Loss at twice the impact of a no-decision. These are the ten "most innocent" pitcher/seasons in my data:
1) Greg Maddux, 1992. Maddux won his first Cy Young Award in 1992, going 20-11 with a 2.18 ERA, four no-decisions. But the 20-11 record significantly understates how well Maddux pitched. Maddux lost 1-0 to the Pirates (April 25), 4-0 to the Reds (May 1), 2-0 to the Giants (May 16; one of the two runs in that game was allowed by the bullpen), 2-0 to the Padres (May 22; complete game), 6-2 to the Giants (May 27; Maddux pitched great but the bullpen was hammered), and 4-0 to Atlanta (July 10; Maddux gave up two runs in eight innings). He gave up four hits and one run in nine innings to Houston on May 11, but the team lost the game in extra innings. Maddux pitched 108.2 innings in the 15 games that he didn’t win that year—more than seven innings per start—and posted a 2.90 ERA in those games.
Maddux was 27.6% responsible for the games he didn’t win that year; the offense, 72.4%. No pitcher in the last 60 years has been less responsible for the games he didn’t win.
I should explain. . . .Justin Verlander isn’t eligible for this category, because there weren’t enough games that he didn’t win. To be eligible for this list, a pitcher to have 25 "points" worth of games that he didn’t win, with a Loss being two points and a No Decision being one. Basically, it takes about 8 to 10 losses to qualify for the list.
2) Gaylord Perry, 1972. A moderately famous season, and another Cy Young Award. Perry went 24-16 with a 1.92 ERA in 342.2 innings. Perry was only 28.5% responsible for the games that he didn’t win that year.
3) Roger Clemens, 2005. Clemens went 13-8 with a 1.87 ERA for Houston.
4) Sam McDowell, 1968. 15-14, 1.81 ERA. This system, if you are trying to decode this, discriminates slightly in favor of a pitcher in a low-run context (1968), and against a pitcher in a high-run environment (2003). But only slightly.
5) Randy Johnson, 2004. 16-14, 2.60 ERA. 290 strikeouts and 44 walks. Johnson, Clemens and McDowell are each 29% responsible for the games they didn’t win, with their offenses being 71%.
6) Tom Seaver, 1967. 16-12, 2.20 ERA.
7) Tom Seaver, 1973. 19-10, 2.08 ERA.
8) Mike Scott, 1986. 18-10, 2.22 ERA, 306 strikeouts. Scott won the Cy Young Award despite poor offensive support, as did Seaver in ’73.
9) Joel Horlen, 1964. 13-9, 1.88 ERA.
10) Steve Carlton, 1969. 17-11, 2.17 ERA.
If you focus on other famous tough-luck seasons—Dave Roberts in 1971, Nolan Ryan in 1987, Bert Blyleven in 1985—they all show up ON the list of pitchers least responsible for their losses, but just not at the top of the list. Nolan Ryan was actually less responsible for the games he didn’t win in 1972, 1973 and 1974 than he was in 1987, although the offense was still 66% responsible for Ryan’s defeats and no-decisions in 1987.
Bob Gibson in 1968 lost 9 games despite a 1.12 ERA. That season doesn’t show up on our list, because Gibson didn’t lose enough games (9 losses, 3 no-decisions equals 21 points; list requires 25), but the offense was 75% responsible for the games Gibby didn’t win that year, even higher than the number for Maddux. If I had used 20 points as the qualifying standard, Gibson—and only Gibson—would have ranked ahead of Maddux in ’92. Other than Gibson, no one with more than two losses would rank ahead of Maddux. Luis Tiant in ’68 would have been very close to the top of the list, had he qualified, and the Deaf Frenchman, Kevin Appier in 1992, would have been on the list, had he qualified. Appier was 15-8 with a 2.46 ERA.
Now, the other end of the scale. The head of the Just Didn’t Pitch Well Brigade was Mark Hendrickson in 2003 (9-9, 5.51 ERA). Hendrickson was 63% responsible for the games that he didn’t win.
Second on the list was Rich Gale in 1979, with regard to which I have a personal memory. Gale was 9-10, 5.65 ERA—but Gale always insisted, more than any other pitcher I ever knew, that he was not responsible for the games that he didn’t win. No matter how badly he had been pitching, Gale would tell you "well, of my last seven games I lost four and had two no-decisions, but I pitched very well in four of those, and gave up only 4 runs in one of the other ones." He was actually famous for doing this, and the people in the press box used to joke about it; you can go back to the Kansas City newspapers at that time, and find places where Gale picks through his losses and writes them all off to bad luck. It turns out that he was near the top of the list of pitchers MOST responsible for his own troubles.
Third, Casey Fossum, 2006 (6-6, 5.33 ERA, 13 no-decisions), fourth, Luis Tiant, 1977, fifth, Doug Drabek, 1998, sixth, Jeff Fassero, 1999, eighth, Colby Lewis, 2003. Colby Lewis in 2003 was 10-9 with a 7.30 ERA. After that he went to Japan for a couple of years, and returned a better man for it, or at least a better pitcher. Ninth, John Doherty, 1993, and 10th, Darren Oliver, 2000.
Career leaders and trailers. . ..in their careers, the pitchers LEAST responsible for the games they didn’t win were:
Rank
|
First
|
Last
|
DPW Pct
|
1
|
Tom
|
Seaver
|
38.7%
|
2
|
Nolan
|
Ryan
|
38.9%
|
3
|
Bob
|
Gibson
|
38.9%
|
4
|
Andy
|
Messersmith
|
39.1%
|
5
|
Mel
|
Stottlemyre
|
39.8%
|
6
|
Joe
|
Horlen
|
39.9%
|
7
|
Jose
|
DeLeon
|
40.0%
|
8
|
Sam
|
McDowell
|
40.0%
|
9
|
Ken
|
Johnson
|
40.0%
|
10
|
Steve
|
Rogers
|
40.0%
|
11
|
Matt
|
Cain
|
40.3%
|
12
|
Pedro
|
Martinez
|
40.3%
|
13
|
Jose
|
Rijo
|
40.3%
|
14
|
Don
|
Wilson
|
40.3%
|
15
|
Gaylord
|
Perry
|
40.4%
|
16
|
Gary
|
Nolan
|
40.4%
|
17
|
Larry
|
Dierker
|
40.5%
|
18
|
Don
|
Drysdale
|
40.5%
|
19
|
Felix
|
Hernandez
|
40.8%
|
20
|
Jon
|
Matlack
|
40.8%
|
The most interesting name there being, perhaps, Jose DeLeon, in that you don’t expect to see Jose DeLeon’s name on a list with Tom Seaver, Nolan Ryan, Bob Gibson and Pedro Martinez. DPW Pct stands for "Didn’t Pitch Well Percentage". The other extreme, the pitchers most responsible for the games they didn’t win, minimum 200 points:
Rank
|
First
|
Last
|
DPW Pct
|
1
|
Darren
|
Oliver
|
52.7%
|
2
|
Aaron
|
Sele
|
51.7%
|
3
|
Kirk
|
Rueter
|
51.0%
|
4
|
Bill
|
Swift
|
50.9%
|
5
|
Steve
|
Avery
|
50.9%
|
6
|
Kenny
|
Rogers
|
50.8%
|
7
|
Russ
|
Ortiz
|
50.6%
|
8
|
Jamey
|
Wright
|
50.6%
|
9
|
Jeff
|
Francis
|
50.2%
|
10
|
Jimmy
|
Haynes
|
50.2%
|
11
|
Brian
|
Anderson
|
50.2%
|
12
|
Sidney
|
Ponson
|
50.0%
|
13
|
Tom
|
Brewer
|
49.9%
|
14
|
Jose
|
Lima
|
49.9%
|
15
|
Jaime
|
Navarro
|
49.7%
|
16
|
David
|
Wells
|
49.7%
|
17
|
Pedro
|
Astacio
|
49.5%
|
18
|
Scott
|
Erickson
|
49.5%
|
19
|
Mike
|
Hampton
|
49.5%
|
20
|
John
|
Thomson
|
49.4%
|