Sorry; failed to post yesterday’s article yesterday.   Let me begin by telling you briefly the stories of two more remarkable long games.

 

May 24, 1929 at Comiskey Park

On May 24, 1929, the Chicago White Sox and the Detroit Tigers squared off in Chicago.   Neither was a good team.  They were both second-division teams, had been in 1928 and would be again in 1929.  On that day, however, both teams had their best pitchers on the mound, Ted Lyons against George Uhle.  Lyons was sited here as perhaps the best pitcher in the American League in 1927.  Uhle, a 27-game winner in 1926, had been injured in 1927 and had struggled through 1928, but had won his first seven starts in 1929—not his first seven decisions, his first seven starts.   He came into the game 7-0 with a 1.43 ERA.   This was his eighth start. 

In this one, however, the White Sox reached him for two runs on two triples in the first inning.  The Tigers answered with three in the second, Uhle himself driving in one of the runs and scoring another one.  Uhle was an outstanding hitter; you probably know that.  He was famous for it.   His career average was .289, the highest of any pitcher ever who had a good long career.

Anyway, 3-2 after two innings.  The Tigers made it 4-2 in the third, another triple being the key hit, a triple by Dale Alexander.   The White Sox got five singles in the fifth inning, three runs out of it, giving them a 5-4 lead.   The Tigers tied it in the top of the 7^th, Charlie Gehringer scoring on a sacrifice fly by Alexander. 

And then the scoring stopped.   No runs were scored in the bottom of the 7^th, or in the 8^th or in the 9^th.   Uhle singled again in the 8^th; otherwise both teams were retired in order. 

Both teams went in order in the 10^th.  There was a single in the 11^th.  In the 12^th the Tigers got runners on first and third on singles by Hall of Famers Charlie Gehringer and Harry Heilmann, but tried a double steal.   The shortstop, covering second, cut off the throw.   Gehringer tried to retreat to third, couldn’t get back, broke to home and was out at home plate.   The threat evaporated. 

Uhle singled again with two out in the 13^th, and Roy Johnson followed with another single, but then Lyons, still pitching for Chicago, got the third out.   In the 14^th Heilmann doubled, but was thrown out at home plate (9-2) trying to score on a single. 

The White Sox got singles in the 14^th and 15^th, but didn’t score either runner.  Still 5-5 through 15 innings.

16^th inning, Roy Johnson led off with a single for Detroit, reached third base but died there.  In the 17^th inning Dale Alexander led off with a single, but rounded first the wrong direction and was nailed off of first base.   In the bottom of the 17^th Ted Lyons led off with a single, but reached only to second base.  Both teams went quietly in the 18^th, and in the 19^th. 

In the top of the 20^th Dale Alexander led off with a single again, the third time in the game he had done so. This time the Tigers pinch ran for him and bunted him to second, but. . .didn’t score.   In the bottom of the 20^th the White Sox got a leadoff single, didn’t score. 

In the 21^st inning Uhle led off with a single, his fourth hit of the game.   Emil Yde pinch ran for Uhle.   He had pitched 20 innings; perhaps that was enough.  Roy Johnson singled; Yde went to third, Johnson to second on the throw.  But then Yde tried to score on a ground ball to second and was out at home.  

Charlie Gehringer hit a fly ball, scoring Roy Johnson, the first run that had scored in 14 innings, giving Detroit a 6-5 lead.   The reliever pitched a clean 21^st inning, and Detroit won the game.   Since the reliever entered the game with a 6-5 lead, Uhle was credited with the win, making him 8-0 after eight starts.  For the game, Uhle pitched 20 innings, giving up 17 hits, 5 runs, 3 walks, 4 strikeouts.   Lyons pitched 21 innings, giving up 24 hits, 6 runs, 2 walks, 4 strikeouts.   The 24 hits allowed included six leadoff singles. 

The 20-inning performance by Uhle came at a high cost.   After the start Uhle was 8-0 with a 1.63 ERA.   He did win his next start, but gave up six runs in that one, then started losing, his ERA ascending every time he even looked at the mound.  He had a 5.94 ERA in June, 5.02 in July, 8.89 in August.   The evidence in regard to Lyons is less convincing, but after posting an ERA just over 3.00 for the first two months, it was 4.40 in June and 4.86 in July.

September 11, 1946

Cincinnati at Brooklyn

It had been eight years since Johnny Vander Meer had pitched his two consecutive no-hitters.  Vander Meer had struggled through the Reds’ two pennant-winning seasons, 1939 and 1940.  He was 5-9 with a bad ERA in 1939, and made only 7 starts in 1940.   He found himself in 1941, led the National League in strikeouts with 202, finished 18-12, 16-13 and 15-16 in a three-year stretch with excellent ERAs of 2.82, 2.43 and 2.87.   He was the #3 starter on those teams, behind Bucky Walters and Elmer Riddle, but he was a really good #3 starter.   He walked too many people, walked 162 men in 1943. 

He was the Army then for a couple of years, came back from the war not quite the same, his strikeout total down by quite a bit.   On September 11 he took the mound with 9 wins, 10 losses, and a 3.64 ERA.

Opposing him for da Bums that day was Hal Gregg.   Gregg had won 18 games in 1945, when the stars were in the Army, but missed a month with an injury early in the 1946 season and was relegated to a spot starter role once he returned.   14,538 fans showed up to watch the afternoon contest.   The Dodgers were still in the pennant race, 3½ games out, although the Reds had been eliminated during spring training. 

Both teams got leadoff walks in the first inning, but neither team got the man around.   Vander Meer started striking people out.  He struck out one in the first, two in the second, one in the third, one in the fifth, one in the sixth.  Here; I’ll build his record out inning by inning:

IP	H	R	ER	BB	SO
1	0	0	0	1	1
2	1	0	0	1	3
3	1	0	0	1	4
4	2	0	0	1	4
5	3	0	0	1	5
6	4	0	0	1	6

 

Hal Gregg, meanwhile, was equally good through four innings:

IP	H	R	ER	BB	SO
1	1	0	0	1	1
2	1	0	0	1	1
3	1	0	0	1	2
4	1	0	0	1	2

 

In the fifth inning Johnny Lukon tripled to left field for the Reds, but tried to score on the play and was thrown out at home plate, Reiser-to-Reese-to the catcher, who was Bruce Edwards. 

Vander Meer retired the side in order in the 7^th and 8^th, gave up an infield single in the 9^th.   Through nine innings these were their line scores:

 

VANDER MEER							GREGG
IP	H	R	ER	BB	SO		IP	H	R	ER	BB	SO
1	0	0	0	1	1		1	1	0	0	1	1
2	1	0	0	1	3		2	1	0	0	1	1
3	1	0	0	1	4		3	1	0	0	1	2
4	2	0	0	1	4		4	1	0	0	1	2
5	3	0	0	1	5		5	2	0	0	1	3
6	4	0	0	1	6		6	3	0	0	1	3
7	4	0	0	1	6		7	4	0	0	1	4
8	4	0	0	1	7		8	5	0	0	1	5
9	5	0	0	1	9		9	5	0	0	1	6

 

And through ten innings:

VANDER MEER							GREGG
IP	H	R	ER	BB	SO		IP	H	R	ER	BB	SO
1	0	0	0	1	1		1	1	0	0	1	1
2	1	0	0	1	3		2	1	0	0	1	1
3	1	0	0	1	4		3	1	0	0	1	2
4	2	0	0	1	4		4	1	0	0	1	2
5	3	0	0	1	5		5	2	0	0	1	3
6	4	0	0	1	6		6	3	0	0	1	3
7	4	0	0	1	6		7	4	0	0	1	4
8	4	0	0	1	7		8	5	0	0	1	5
9	5	0	0	1	9		9	5	0	0	1	6
10	5	0	0	1	11		10	5	0	0	2	6

Hugh Casey replaced Hal Gregg in the 11^th inning, pitching for the Dodgers, but Vander Meer was on another strikeout run.   He struck out two batters in the 9^th, two more in the 10^th, two more in the 11^th.   

In the thirteenth inning the Dodgers got a runner in scoring position with two out on an error, a failed bunt, and then a successful sacrifice bunt.  It was the first time the Dodgers had had a runner in scoring position since the first inning.   Weren’t able to score.

Cookie Lavagetto doubled for the Dodgers with one out in the 15^th inning, and Bruce Edwards was intentionally walked.  That was the first walk the normally wild Vander Meer had issued since walking the leadoff batter of the game, Eddie Stanky.  But Vander Meer retired the next two batters. 

He was done for the day, then.   Completing his log:

 

VANDER MEER
IP	H	R	ER	BB	SO
1	0	0	0	1	1
2	1	0	0	1	3
3	1	0	0	1	4
4	2	0	0	1	4
5	3	0	0	1	5
6	4	0	0	1	6
7	4	0	0	1	6
8	4	0	0	1	7
9	5	0	0	1	9
10	5	0	0	1	11
11	6	0	0	1	13
12	6	0	0	1	13
13	6	0	0	1	13
14	6	0	0	1	14
15	7	0	0	2	14

 

Fifteen innings, fourteen strikeouts, no runs.  It was not Vander Meer’s last hurrah; he would win 17 games in 1948, and pitch three shutouts in 1949.

The nothing-nothing contest went on to the 16^th inning.  Ad Gumbert replaced Vander Meer.  He was equally effective.  In the 16^th inning a Dodger reached on an error; in the 17^th, a single, and in the 18^th, nothing.   Still 0-0. 

Dain Clay of the Reds, who had started the game by drawing a walk, started the 19^th inning by drawing another one.  A bunt moved him to second base,  followed by an intentional walk to set up a possible DP.   Bert Haas singled to right field.  Dixie Walker raced in and threw home.

Got him.   Dain Clay was out at the plate.

Ad Gumbert retired the side in order in the bottom of the 19^th. 

The game was called on account of darkness.  Did Ebbets Field not have lights?  The game ended in a nothing-nothing tie after 19 innings. 

Well, you remember that Johnny Vander Meer’s second no-hitter in 1938 was the first night game played at Ebbets Field.  You remember?

My understanding is that the rule at that time was that a game which began without lights had to be completed without lights.   You couldn’t turn the lights on in the middle of the game.  There was a fear that converting a game from a day game to a night game in the middle of the game would give an unfair advantage to one team. 

I know that was the rule in 1941, and I know it was changed within a few years, but I don’t know exactly when.  The same day that this game was played, the Cubs and Braves also played to a 17-inning, 3-3 tie in Boston, that game also called on account of darkness.  That game was 3-3 after five innings, and then the teams played scoreless baseball for 12 more innings. 

 

 

Game Scores, Version Two

 

  Before we get to the next installment, Tom Tango sent me an e-mail comparing his other version of Game Score for the Vern Law game to my Game Score.  My Game Score has the game at 118; his version would have it at 126:

40 points as a base or starting point,

+2 points for each out recorded (54 outs, +108, makes 148),

+1 point for a strikeout (12 strikeouts, +12, makes 160),

Minus 3 for each run allowed, earned or unearned (2 runs, -6, makes 154),

Minus 2 for each walk (2 walks, -4, makes 150),

Minus 2 for each hit (9 hits, -18, makes 132),

Minus 6 for a home run allowed (1 homer, -6, makes 126).

 

 

What he was wondering about, actually, is "Which games, within your data, would have the largest separation between your Game Score, or "original" Game Score, and his modification of it.   I’ll get to that issue in a minute here, but first a few other questions I must address, in my fanatical and borderline anal-retentive manner:

1)      In my version, the Vern Law game is the best outing by a starting pitcher among the 329,988 games, scoring at 118.   By Tom’s version, however, the best game is the Herb Pennock game that I also wrote about earlier in the series, the 15-inning shutout in 1925.   That scores at 127 (Tom’s version) versus 114 in the original version.

 

2)     There are four games which score at 126 in his version—the Vern Law game, the well-known start by Tom Cheney in 1962, in which he struck out 21 men in 16 innings, a start by Dean Chance against the Yankees in 1964, in which he pitched 14 shutout innings with 3 hits and 12 strikeouts, and the Johnny Vander Meer start that I wrote about above.  September 11, 1946.

 

3)     In my version, the worst start was the Hod Lisenbee game, which was September 11, 1936; I wrote about it.  That scores at -35 in my version, -34 in Tom’s.   In Tom’s version it is tied for the worst game in the study with a start by Jeremy Guthrie on May 25, 2015.  Pitching for the eventual World Series champion Royals, Guthrie gave up 8 runs in the first inning.   Sent back to the post for the second, he gave up a 3-run homer with no one out, leaving the game then having pitched one inning, allowing 9 hits, 11 runs, all earned, and four home runs.   That also scores at -34 in Tom’s version of the Game Score. 

 

4)     In my version of Game Score, which I invented in the 1980s, the average Game Score for all 329,988 games is 50.4   In Tom’s version, the average Game Score is 51.2.   In his version the average score would drop to almost exactly 50 if he were to increase the penalty for a Home Run Allowed from 6 to 8, although, for reasons I will explain later, I think it would be a mistake for him to do that. 

 

5)     In my version, the standard deviation of Game Scores is 17.57.  In Tom’s version it is 20.24. 

 

It is intended to be a 0-to-100 scale.    My system, however, has 55 starts in which the Game Score exceeds 100, and 137 starts with a Game Score less than zero.   192 total starts which are "out of bounds", or one game in 1,700, more or less. 

 

Tom’s version, on the other hand, has 1,409 games which are "out of bounds"—327 games over 100, and 1,082 which have scores less than zero.    It is one game in 234.  

 

6)     I sorted the 329,988 games into three groups by each method—

a)     The top 100,000 games

b)     The middle 129,988 games

c)      The bottom 100,000 games.

 

My system—the original system—has a very small advantage in terms of pushing the best games to the top and the worst games to the bottom.  Again, the output of the systems is very similar, and you would have to study very large groups of games to find the differences, but. .

 

In my system, the top 100,000 pitchers (with the second-sort being the score in Tom’s rating system) . . . .start over.   The top 100,000 pitcher/games in my system had an .854 winning percentage, a 1.19 ERA, and allowed 1.40 total runs per nine innings (earned and un-earned.)  Their teams had a winning percentage of .7935. 

All of those numbers are slightly worse for the top 100,000 in Tom’s system (with the second sort being, of course, the score in my rating system.)  In his system the top 100,000 pitcher/games had an .850 winning percentage, a 1.24 ERA, and allowed 1.43 total runs.  Their teams had a winning percentage of .7916.

 

When you focus on the BOTTOM group, the worst starts. . .the worst starts in my system are worse than those in his.  In my system, the 100,000 worst starts produced an .080 winning percentage, a 10.18 ERA and 11.19 total runs per 27 outs.  Their teams had a .222 winning percentage.  The worst starts in Tom’s system are not quite as bad:  .081 winning percentage, 10.15 ERA, 11.14 total runs per 9 innings, and a .226 team winning percentage. 

 

In his system, the 100,000 best starters pitch a few more innings—24.4 per start versus 24.3.   His 100,000 best starts have a .005 advantage in innings pitched.  Of more significance, his preferred starters have quite significantly fewer walks, 4.2% fewer per inning, because he weights a walk at -2 and I weight it at -1.   And his top 100,000 starts have far, far fewer home runs allowed—27% fewer. 

They have 27% fewer home runs allowed, but 3% more hits allowed, 7% more doubles allowed, and 11% more triples allowed.   My "good" starting pitchers averaged 5.80 strikeouts per nine innings; his, only 5.69.  I don’t quite understand that difference, since both systems weight a strikeout the same. 

 

            I think my systems "wins" this comparison, basically; it does a very slightly better job of identifying winning pitchers and pitchers with low ERAs, and does a much better job of keeping the scores in the zero to 100 range.  The run support for my top 100,000 starts is very, very slightly less than for Tom’s 100,000.  But I have a suggestion as to how his system might move ahead on these comparisons.  

            The average pitcher in this data allows .6003 home runs per start; let’s call it .6.  Weighting those by 6 points each, there is an average deduction of 3.6 points per start for home runs allowed, in Tom’s system.

            That seems to me to be a rather large weight to assign to a home run allowed.  If, instead of weighting home runs at 6 each, he weighted extra base hits allowed at 2 points each. ..there are 1.91 extra base hits allowed per start.   If he weighted extra base hits at 2 each that would be an average deduction of 3.82, which would reduce the average score in his system from 51.2 to 51.0.  I am pretty sure that it would also reduce the standard deviation of scores in his system, keeping more of the scores within the zero to 100 range.  And I would GUESS that this would also tend to improve the performance of the system in terms of pushing the best games to the top.

            Well, hell, now that I have come this far, I might as well experiment with the system and see what I have. 

            I ran scores for a "third version", which was Tom’s Game Score system, except that I deducted 2 points for each extra base hit, rather than 6 for a home run allowed, and also I deducted 2 points for a hit batter, consistent with Tom’s practice of deducting 2 for a walk. 

            When I did that, the average Game Score for the third system dropped from 51.2 to 50.7, and the standard deviation dropped from 20.24 to 19.66.   Those numbers are better than Tom’s current system, but still behind the original system (50.4 and 17.57).   But this revision did NOT improve the system in terms of identifying games with a high winning percentage and a low ERA.   The winning percentage of the top 100,000 games stayed at .850, the team winning percentage stayed at .792, and the ERA increased from 1.24 to 1.26. 

            By the third system, the Vern Law game moves back to being the highest-scoring game in the data. 

 

7)     We get, finally, to the question which Tom actually posed, which is:  what games have the largest discrepancies between the two Game Scores?

 

The largest discrepancy between the two Game Scores is 32 points, for a recent start.    It was the only start of his career for a Milwaukee  pitcher named Michael Blazek, and. . .well, he was Blazecked.  He was Blazecked hard.  He was facing off against some guy named Max Scherzer, and Bryce Harper homered against him in the first inning and again in the third.  That wouldn’t have been too bad; Bryce Harper is Bryce Harper, after all, but Blazek also gave up home runs to Ryan Zimmerman, Brian Goodwin, Wilmer Difo and Anthony Rendon.  And he only pitched two and a third innings.   It was not a good outing. 

 

It was a bad outing by any standard, but that is not actually the point here.  The point here is that the Game Score for that game is 14 on my scale, the original Game Score, and -18 on Tom’s scale.   There is a 32-point separation between the two.  That is the largest for any game in my data. 

 

There are three games in which there is a 31-point separation between my Game Score and Tom’s.   Those three starts were

·      By Bill Kerksieck (Phillies) on August 13, 1939,

·      by Al Milnar (Cleveland) on August 22, 1940,

·      and by Bill Krueger (Oakland) on August 2, 1983.

 

All three of those starts score at +17 by my system, but at -14 by Tom’s system.  That’s a 31-point discrepancy.  There are three additional games in which Tom’s system gives a score 30 points worse than mine, and there are four games in which his system gives a score 29 points worse than mine. 

 

On the other side we have games in which Tom’s system gives a score BETTER than my system.  On that end of the spectrum, the #1 and #2 games ever both occurred, remarkably enough, in the same game.  It was that Uhle/Lyons game that I wrote about earlier, May 24, 1929.  Remarkable game; that was why I wrote about it. 

In that game, Ted Lyons’ performance scores at 77 on my Game Score, but at 100 on Tom’s.  Lyons pitched 21 innings, giving up 24 hits and 6 runs—not really your typical pitcher’s box score line.  There is a 23-point discrepancy between the evaluation of the game by the two systems. 

And right behind it—George Uhle, in the same game.   Uhle pitched 20 innings, giving up 17 hits, 5 runs.   By my system, the game scores at 89.   By Tom’s system, it’s 109—a 20-point discrepancy.

Behind that, there are two games in which his system scores 18 points higher than mine, one game in which he is +17, and five games in which he is +16.   To generalize, my system gives a significantly higher score if the pitcher gives up a LOT of home runs, and his system gives a significantly higher score if the pitcher pitches a long game, extra innings, and gives up multiple runs but no home runs. 

 

On average, the absolute difference between the two systems is 4.33 points.  There are 22,399 games in my data in which the two systems get exactly the same Game Score, and there are an additional 44,030 games in which the difference between the two systems is only one point.