Remarkable stuff I didn’t know department:   Robin Roberts in 1952 won 28 games.   No National League pitcher since then has won as many, but here’s what I didn’t know:   Roberts went almost a month without winning.    Roberts started and won on May 21 and on June 20; between those dates he made six straight starts without getting a win.   The rest of the season he won 28 out of 31 starts.

Also, Roberts in 1952 won 28 games despite making only 37 starts (all of his wins were as a starter, although he did pitch twice in relief.)  That actually was fewest starts that Roberts made in any season between 1950 and 1955. …39, 39, 37, 41, 38, 38.   So he made the fewest starts that he made in a six-year stretch, didn’t win for almost a month—and still came close to winning 30 games.

* * * * * * *

Joe DiMaggio won the American League MVP Award in 1939 although he missed 32 games with injuries.   The Yankees won-lost record in the 32 games that the MVP didn’t play was 27-5. 

* * * * * * *

Here’s something else I didn’t know;   I had no idea how well Don Larsen had been pitching in late 1956, prior to his perfect game.    On June 9, 1956, Larsen’s ERA was 5.57.   Although he pitched very well in June and July, as of August 6 his ERA was still 4.23.

From then to the end of the season—not counting the World Series—he pitched 73.1 innings in 11 games with an ERA of 1.79, cutting his final ERA to 3.26. 

* * * * * * *

It is within human nature, I think, to become less interested in fiction as we age.  I’m 62; I don’t read much fiction any more.  I find that a lot of people in my age cadre don’t.  Some do; my sister, who is 72 and has limited mobility due to a stroke, still reads a couple of novels a week and has for 60 years; it becomes increasingly difficult to recognize the stuff she has already read.

But a lot of us lose interest in fiction as we age, and by "fiction" I mean to include movies and sit coms and stuff that tells made-up stories.   When we were young we would watch anything on television.   The entire nation in the 1960s watched shows like "Bewitched" and "Gilligan’s Island"—shows that seem now preposterously simplistic. They had audiences in their day in the tens of millions. Green Acres.  Grown-ups watched "Green Acres" and "Hogan’s Heroes" and laughed at the jokes.

I can pinpoint a few moments in my life at which this pleasure fled from me.   When I got out of the Army in 1973 I didn’t want to do anything except get myself a room and lie on the bed and watch television.  In the Army people tell you what to do about 120 hours a week.   Some people are comfortable with that; some people stay in the Army because they need somebody else to structure their lives. Some people get out of prison and have to get back in because they can’t live unstructured lives. I’d have killed myself if I couldn’t get out of the Army after a couple of years.   Having somebody tell me what to do all the time was hell to me.  I just wanted to NOT DO ANYTHING for a while.

I got myself a cheap room and a little black-and-white TV, which by the way still works; this was 38 years ago. It took me about two weeks before I was ready to get up and engage the world again, but it didn’t take nearly that long for the TV to grow tiresome.  Between 1971 and 1973, in my view, television had turned to mush.  I couldn’t believe how bad the shows were.   CBS had cancelled, in the words of Pat Buttram, "every show that had a tree in it."  It must have been. .. .oh, I don’t know; November 5, 1973.  I’ve never felt the same about television since.

When my wife and I were first married there were shows we would watch, fictional shows; we’d watch Barney Miller and Cheers, a few others.   I can pinpoint the moment at which they lost us. We used to watch the show LA Law.  About the third year of LA Law there was an episode in which Harry Hamlin, who played a lawyer, was cross-examining a policeman during a deposition about a shooting, and the cop was saying, "What would you do, if you were in this situation, where a punk was pointing a gun in your face and you had to decide in the matter of a second whether he was going to live or you were, what would you do?"

"I don’t know," said Hamlin, "I’ve never faced that situation."

"Well, face it now," says the cop, and he pulls his gun and points it in Hamlin’s face.  It was a very dramatic moment, and we gasped, and then they cut to a commercial.

And when they came back from the commercial, they had just dropped the whole thing; the cop had put his weapon away, and the deposition had ended and the show had moved on.   Their message to their viewers was, "We control this experience; we can manipulate you to think what we want you to think, and then we can move on however we want to move on."   That’s right, you can—but we’re done.   We lost interest in the show after that, because. …well, it was just a show.    They had demonstrated too forcefully that they could do whatever they wanted to with the show.   Without parameters it was just play-acting.

In a sense the show had jumped the shark; we have that expression for that now, which we didn’t then.   In a sense the show had jumped the shark, but the transition was more profound than that.   It wasn’t that the show had changed but that we had changed; we had surrendered our disbelief for the last time.   We had gotten too old for television; I guess I was 35.

Sometimes old people talk about how things aren’t the way they used to be; politics used to be civil, and marriage used to be a real commitment, and things used to be built to last, and athletes used to care about their fans, and businessmen used to be honest, and people didn’t file lawsuits because their coffee was too hot, and things used to be made in America, and we didn’t used to have beggars all over the streets, and TV shows used to have trees in them. Some of this is real; the world does change, and some things aren’t what they once were, and some of the changes are for the worse, although more of them are for the better.  Television grows constantly more sophisticated in an effort to hold on to an ever-evaporating audience.  A lot of it is just. ..well, stuff wears out on you.   The magic doesn’t work anymore. I am not any less interested in the world than I used to be, but as I grow older I am more interested in the world that really is, rather than in the stuff that people make up to try to entertain me.

Snakebite Games

For reasons having nothing to do with anything I was interested in a game that Herm Wehmeier pitched on July 18, 1951.  Wehmeier pitched 7 innings, giving up only 3 hits and 1 run, which was un-earned, but was charged with a loss as the Reds were shut out.  The official stats show it as an earned run, but if you go back to the game accounts, all published accounts of the game highlight the fact that the runner who scored reached as a result of an error, and that the run was un-earned.   

Anyway, I got interested in "Snakebite Games", defining a Snakebite game as any game in which a starting pitcher

a)      Does not give up an Earned Run, and

b)      Has a Game Score of 60 or higher, but

c)      Is charged with a loss. 

By the way, Luis Tiant had a game against the Rangers (May 8, 1976) in which he did not give up an Earned Run but had a Game Score of 23—5 innings, 6 hits, 1 strikeout, 3 walks, 9 runs allowed, but all 9 of them un-earned.   Bobby Jones of the Mets also gave up 9 runs, all un-earned, on May 2, 1994, and Chris Reitsma did it in 2002.

The king of all Snakebite games, of course, was the Harvey Haddix game (May 26, 1959) in which Haddix pitched 12 perfect innings, but lost the game on an un-earned run in the 13^th; I’m assuming you all know that story, and probably many of you remember the game on April 23, 1964, in which Ken Johnson of the Houston Colts pitched a no-hitter, but lost on an un-earned run.  But here’s some more you don’t remember…

On April 29, 1953, Curt Simmons of the Phillies shut out the St. Louis Cardinals for 10 innings on 2 hits, but lost the game in the 11^th inning after a player named Jack Lohrke made an error to start the 11^th inning.

On September 24, 1954, Hal Brown of Baltimore shut out the Red Sox in Fenway Park for 10 innings, lost in the 11^th inning on an error by shortstop Milt Bolling.   Brown had three Snakebite losses in his career.

On August 10, 1955, Frank Sullivan gave up two runs to the Yankees in the third inning on a walk, an error by the second baseman, two bunts, and a single.  Sullivan then shut out the Yankees for the next nine innings; the game was tied 2-2 in the bottom of the 13^th, Sullivan still pitching.   In the bottom of the 13^th the Yankees scored again on an error by the shortstop, a bunt, and a two-out single.

Don Larsen was the winning pitcher in that game; both Larson and Sullivan went all 13 innings.  In 1956—two months before his perfect game--Larsen LOST to the Red Sox after he pitched 10 shutout innings when the Yankees made two consecutive errors to open the bottom of the 11^th.   This was the turnaround game for Larsen that started the run that culminated in his perfect game; that’s how I happened to notice that.

On May 26, 1957, Bud Daley shut out the White Sox through 9 innings, lost on an un-earned run in the 10^th inning.      

On June 3, 1964, Don Drysdale shut out the Phillies for 10 innings, striking out 10, lost on an un-earned run in the 11^th on a ground ball to Junior Gilliam that should have ended the 11^th inning. 

On July 6, 1968, Sudden Sam McDowell pitched a 3-hitter and struck out 14 batters, walking only 2, but lost 2-0 on an error by the first baseman and a passed ball by the catcher.

On May 26, 1975 (the same date as Daley’s game, and the 16^th anniversary of Harvey Haddix’ game) Ed Halicki shut out the Phillies for 10 and two-thirds innings, giving up only 4 hits, striking out 10, walking 1.    With two out in the bottom of the 11^th Larry Bowa singled, was picked off first but went to second when the throw got away from the first baseman, and then scored on another error.

On June 16, 1986, Charlie Hough carried a no-hitter and a 1-0 lead into the bottom of the ninth inning in a game in Anaheim.  In the bottom of the 9^th Bobby Valentine put George Wright, who was a very good outfielder, in left field as a defensive replacement for Gary Ward.   Hough struck out the leadoff hitter, but then Jack Howell, pinch hitting for Gary Pettis, hit a drive to left that Wright misplayed into a three-base error, putting the tying run on third base with one out.   Wally Joyner then singled, tying the score at 1-1—the Angels only hit of the game.  Joyner took second on a Passed Ball. With two out, George Hendrick at the plate, Joyner was moving from second on a 3-2 pitch.  Hendrick struck out but the ball got away from the catcher. Hough failed to cover the plate and Joyner scored from second with the winning run.  

On April 15, 1989, Pete Smith of the Braves pitched the game of his life at Candlestick Park against the Giants, 1989 National League pennant winners with Kevin Mitchell and Will Clark.    Smith gave up only 3 hits, struck out 13, walked one, but lost the game 1-0 due to an error by Dale Murphy.  

The two worst snakebite games of the 1990s were both pitched by Kevin Tapani of the Twins, and occurred on June 27, 1991 and June 27, 1994.  On that date in 1991 Tapani limited the Blue Jays to four hits, walked no one and struck out 7, but lost the game 1-0 when Shane Mack failed to catch a two-out fly ball. Three years later it was Déjà vu all over again; Tapani throttled the Royals on five hits, no walks, 8 strikeouts, but lost the game when center fielder Alex Cole set up a 2-run fifth inning with an error.  

With the passage of time both the number and the severity of snakebite games has diminished sharply. I find no record of any game since 1990 comparable to Smith’s or Charlie Hough’s. In my data I have 124 snakebite games in the 1960s, 150 in the 1970s, 119 in the 1980s, 93 in the 1990s, only 68 in the years 2000 to 2009. Most of the decline is due to a reduction in the number of un-earned runs.

Pete Broberg on May 3 and May 10, 1972, is the only pitcher in the last 50 years to have lost snakebite games in consecutive starts.  The next month, June of 1972, Scipio Spinks lost two in three starts, both of them to the Dodgers.  Nolan Ryan lost two in three starts in July, 1984. 

Don Drysdale in 1964 is the only pitcher in the last 50 years to have lost 3 snakebite games in a season.                

Tommy John was snakebit six times in his career, the most of any pitcher since 1952:

Tipping

Here’s a new one.   For Christmas I usually buy my wife a box of good chocolate truffles, box of four costs $10.  So I buy the truffles, and the guy who puts them in the box seems to be making quite a show out of that and I’m thinking this could go a little faster, but we eventually get to the register and I whip out my Mastercard, and. ..here’s a space for the tip.   $10 isn’t enough for four chocolates; now you’re supposed to tip for putting them in a box as well.

Missing DiMaggio

One useful thing about Joe DiMaggio’s career is that, while he had many injuries and usually missed some games every year with something, he pinch hit only 14 times in his career.   This makes his career useful for doing certain types of analysis that you couldn’t do with a lot of partial games.  

So I got to wondering what the Yankees’ record was with Joe DiMaggio in the lineup and with DiMaggio out of the lineup.   Actually that isn’t exactly what I was wondering; exactly what I was wondering is whether I could accurately predict that. . . .predict it after the fact, of course, but predict it, not knowing the answers.   It’s a way of testing whether the things we think we know are useful and practical, or whether real life is so complicated that the things we are so proud of knowing don’t actually explain anything.  

How do we figure this? 

Let’s start in 1936, which was DiMaggio’s rookie season. DiMaggio missed 17 games with injuries—the first 17 games, incidentally.  DiMaggio missed 17 games in 1936, which were the first 17 games, 6 games in 1937, which were the first 6 games, and 12 games in 1938, which were the first 12 games. This creates other issues. Anyway, the first thing we have to estimate is how many runs the Yankees lost when DiMaggio was out with injuries.

Suppose we assume:

1)       That DiMaggio created 128 runs in 1936, making 434 outs,

2)      That the league run environment, park-adjusted to Yankee Stadium, was 5.18 runs per game,

3)      That 26 outs are a game (one out per game tends to disappear in the statistical mist)

4)      That a replacement level center fielder would create runs at a level 82% of the league norm.

None of these are truly "assumptions", of course; we don’t ASSUME that DiMaggio created 128 runs in 1936, we take that number from a source, which is the Bill James All-Time Handbook.   If it quiets your apprehensions to know this, it makes NO difference in the final accounting if you assume that a game is 27 outs, rather than 26, or that the replacement level is 84% of the league norm, or 90%; it makes no difference.   It makes a theoretical difference but not a practical one in this case.

Anyway, DiMaggio created .295 runs per out, whereas the replacement level is .164 runs per out.  Every out that goes to someone else while DiMaggio is out of the lineup costs the Yankees about .131 runs.  Since DiMaggio was out for 17 games and made 3.14 outs per game when he was in the lineup, he was absent from the lineup for about 53 outs.   Since the Yankees lost .131 runs per out while he was missing, the Yankees probably lost 7 runs (offensively) due to DiMaggio’s absence.  

The Yankees scored 1,065 runs in 155 games. With the addition of those 7 runs they lost while DiMaggio was out they would have scored 1,072, or 6.92 runs per game. We assume, then, that while DiMaggio is IN the lineup the Yankees should have scored 6.92 runs per game, whereas when DiMaggio was out of the lineup they should have scored only 6.50 runs per game.  

That covers the offense, but what about defense?   Here my method is so crude that it hardly merits explanation, but. …I don’t have a week to put into this.   We estimated in the Win Shares method that 16% of DiMaggio’s career value was in his fielding, which means that 84% was in his hitting.  That also means that his defensive value was 19% of his offensive value—over the course of his career.

Over the course of his career, but what about in 1936?   I chose to assume that a player’s defensive value relative to his offensive value decreases constantly over his career, and that his defensive value at age 40 would be zero. What, then, would be his defensive value at age 21?

It’s 31% of his offensive value. ..don’t ask me to explain why.  Subtract his age from 40, multiply by 1.64, divide by 100; over the course of his career it works out to 19%.  At age 21 it works out to 31%.   

If the offensive loss is 7 runs (actually, it was 7.03 runs) and the defensive loss was 31% of that, then the anticipated defensive cost of DiMaggio’s outage is about 2.2 runs, 2.19 runs.    The Yankees allowed 731 runs in 1936.   That means that had DiMaggio not missed any games, they should have allowed about 729 runs in 155 games, or 4.70 runs per game.   When DiMaggio was hurt, this should have increased to about 4.83 runs per game.   

We can now guesstimate that when DiMaggio was in the lineup, the Yankees should have scored about 6.92 runs per game and allowed 4.70, which means they should have had a winning percentage of about .684 with DiMaggio in the lineup, and (skipping a couple of steps, which you can easily re-construct) of .644 with DiMaggio in the lineup.   

            However, had they played .684 baseball with DiMaggio in the lineup and .644 without him, the Yankees’ record in 155 games should have been 105-50.  In fact, their record was 102-51.  If we assume that their won-lost underperformance was the same with and without DiMaggio, that means that the Yankees’ record should have been about 91-45 with DiMaggio in the lineup, and 11-6 without him.

            And, in fact, it was; the 1936 Yankees in fact were 91-45 with DiMaggio in the lineup and 11-6 without him.  Of course, it doesn’t always work that way, and we’ll get to those details in a minute, but let’s parcel out the data, rather than just dumping it on your head.  First, here is some basic data to establish a frame of reference:

First, how many runs did the Yankees probably lose, per year, when DiMaggio was out?   Seven in 1936; as many as 48 in 1949:

The 1936 Yankees could have expected to score 6.92 runs per game with DiMaggio in the lineup.   Over the course of DiMaggio’s career run scoring levels decreased sharply, and the Yankee offense never reached that level again:

            The defensive loss we will spin off of the offensive loss:

And from this, we can estimate the Yankees’ expected winning percentage each season with DiMaggio in the lineup and without him:

            The 1940-1941 Yankees were not a lot more than a .500 team without DiMaggio, and in 1941 his presence in the lineup probably improved the expected winning percentage of the team by a whopping 83 points.  This figure was decreasing steadily over the next decade, but in 1951 DiMaggio was still a very good player, still probably improving the expected winning percentage of the team by about 20 points.  Those are the figures that aren’t adjusted for the real performance of the team. When we adjust for the team’s actual won-lost record, we can get the expected won-lost record with and without DiMaggio:

That’s the theory. In fact, the Yankees’ record without DiMaggio in the lineup was 171-106, or six and a half games better than we had expected.  We had expected that the Yankees’ winning percentage without DiMaggio would drop from .644 to .595; in fact, it dropped only from .640 to .617.

Basically all of the discrepancy was in 1939.   In 1936 we had expected the Yankees to go 11-6 without DiMaggio, and they did.  In 1937 we had projected 4-2 without DiMaggio; the actual won-lost record without DiMaggio was 4-2.   In 1938 we had expected them to be 7-5 without DiMaggio; they were actually a game worse than that, at 6-6.

In 1939, as noted early in the piece, the Yankees should have been 21-11 without DiMaggio in the lineup, which is still actually really good—but in fact, they were 27-5, six games better than really good.

In 1940 we had expected them to be 12-11 without DiMaggio; they were actually 10-13.

In 1941, the year of the streak, we had expected them to be 10-7 without DiMaggio; they were actually 11-6.

In 1942 DiMaggio didn’t miss any games. 

In 1946 we had expected the Yankees to be 12-10 without DiMaggio; they were actually 11-11.

In 1947 we had expected them to be 8-6 without DiMaggio, and they were 8-6.

In 1948 we had expected them to be 1-0 without DiMaggio; they were 0-1.

In 1949, when DiMaggio missed half the season with a heel injury, we had expected them to be 47-31 without DiMaggio; they were actually 47-32.  

In 1950 we had projected them to be 10-6 without DiMaggio, and they were 10-6.

In 1951 we had projected them to be 24-14 without DiMaggio in the lineup, but they were actually 26-12.  

In his career, we had expected that in the 277 games that DiMaggio missed with injuries, the Yankees would score 1,428 runs and allow 1,148.   In fact, they scored 65 runs more than that (1,493) and allowed 42 less (1,106).  They exceeded expectations with DiMaggio out by much more in runs than they did in wins, but under-achieved in the won-lost record.  The Yankees actually allowed fewer runs per game with DiMaggio out of the lineup (3.99) than with DiMaggio in the lineup (4.07), which proves that DiMaggio was actually a bad defensive player.   Or not.

It might go without saying both that the deviation from expectation is well within the range of chance (the random chance that a 164-112 team would go 171-106 is 24%) and that the method is too crude to even speculate as to whether the deviations are in fact the result of chance, or result from the shortcuts and inherent defects of the method.  The method has limited ability to generalize to other players.  You couldn’t apply this method to Cal Ripken, for example, because he never missed any games. You couldn’t apply it profitably to Stan Musial or Billy Williams or Ron Santo because they missed so few games that you’d have only the theory, with no reality check.

You couldn’t really apply the method to Ted Williams, because Williams pinch hit 144 times in his career and had irregular absences that would undercut the plan.   A superstar who is setting out the game with a marginal injury will pinch hit when the team is behind, but stay on the bench when they are ahead.   This makes the group of games in which he plays inherently different than the games in which he doesn’t play, and this queers the data.    Throwing out the games in which he pinch hit doesn’t help; they’re still unequal groups.  Irregular absence with "nagging" type injuries are an extension of the problem. When a superstar is being bothered by a nagging injury, he tends to play against the other contenders, and to sit out against the second-division teams, or tends to play when he has the platoon edge but sit down when he doesn’t.

This isn’t an issue with DiMaggio because DiMaggio has "clean" injuries; he plays every game when he is not hurt, and then he’s out of the lineup regardless of who the Yankees are playing.   This makes the DiMaggio data a truer test case than almost any other player, probably any other player—and there are a dozen problems I haven’t dealt with in making it work with DiMaggio.

Still, there is something that I like about the concept.   Our method concludes that the 1941 Yankees were 83 points better (.642 to .559) with DiMaggio than without him.  That would be a useful piece of information to have in an MVP debate, wouldn’t it—what the winning percentage of each team should be with and without each player?  I think it would be.  But I don’t see how we could ever establish that the method does in fact work, given the complications of real-life data.