BILL JAMES ONLINE

The Better League, 1

By Bill James
June 6, 2022
                                                              The Better League, 1

World Series and All Star Games

            

            Back when the two major leagues were truly different entities with competing sets of fans, back when men were men and our old LaSalle ran great except that new tires only lasted a couple of hundred miles, back in the good old days there were passionate American League and National League fans, who would argue about which was the superior league and really care about the subject.  The two pieces of evidence universally accepted in this debate were the World Series results and the All Star Games.  

            The World Series is just seven games a year and the All Star Game just one, so it is not irrational to argue that that isn’t enough evidence to be meaningful, but. . .that’s what we had.  I’m suggesting here a way to systematically evaluate what we have; I’m not suggesting that it is 100% reliable, 90% reliable, or 10% reliable; it is just what it is.

            OK, in 1979 the National League team won the World Series, 4 games to 3, and the National League won the All Star Game.   I record that as 7 to 4.  The National League gets:

 

            One point for each World Series Win (4), plus,

            Two points for winning the All Star Game (6), plus

            One point which every league gets every year (7).

 

            The American League one point for each World Series win (3), plus one point which every league gets every year (4). 

            Two questions there:   Why does an All Star Win count two points, rather than one, and why does every league get a point every year?

            The All Star game is easy to dismiss as "it’s just one game", which was the argument I used for the first 25 years of my life, when I was an American League fan and we always lost the All Star game, but.. . .where was I?   It’s just one game, but the thing is, the team that won last year’s game will win this year’s game again more than two-thirds of the time (67.4%).   Does that seem random to you?  "Random" means that whoever won last year has a 50% chance to win again this year.  

            The history of the All Star Game is one of long streaks of domination by one league or the other.   The American League won 12 out of the first 16, then the National League won 33 out of 41, then the American League won 26 out of 32.   The history of the All Star game isn’t a coin flip every year; it’s dominance by one league or the other.  That makes it harder to say that it’s just one game.

            One game in the World Series represents ONE TEAM from the league.  The American League was almost certainly stronger than the National League by 1907, but the National League won the World Series in 1907, 1908 and 1909.  It’s not really surprising, because it is one team from the league, and the NL had three terrific teams.  They had three teams that won 100 games every year because the other five teams were terrible.   Anyway, that’s the best I can do; I’ m counting the All Star games as two wins each.

            And why do we add one win and one loss to every league every year?  

            Because zeroes and ones are destructive in statistical analysis.  Some leagues (the Federal League, 1914-1915, the Player’s League, 1890). . . some leagues there is no data.  We don’t want them resting at zero.  You can’t logically ASSUME that teams that have no data are inferior leagues.   Giving them 1-and-1 plants them at .500.   Putting them at .500 relative to other leagues may be overly generous, but that isn’t actually a problem, because all of those leagues have obvious indications that they are inferior leagues, and will be buried in standings by those other indications anyway.  

            So the National League in 1979 is 7 and 4, .636.  We put this now into an eleven-year moving average, weighted toward the center:

 

1974

NL

1

4

1

0

4

5

1975

NL

4

3

1

0

7

4

1976

NL

4

0

1

0

7

1

1977

NL

2

4

1

0

5

5

1978

NL

2

4

1

0

5

5

1979

NL

4

3

1

0

7

4

1980

NL

4

2

1

0

7

3

1981

NL

4

2

1

0

7

3

1982

NL

4

3

1

0

7

4

1983

NL

1

4

0

1

2

7

1984

NL

1

4

1

0

4

5

 

            The top line of that chart; in 1974 the NL team lost the World Series, 4 games to 1, but he NL won the All Star Game, so that’s 3 and 4; add 1 and 1 and it is 4 and 5.   The 1974 season is entered into the data as 5 wins for the American League, 4 for the National League. 

            Then we weight the data, as follows:

            The focus season (1979) is weighted at 10. 

            The season before and the season after (1978 and 1980) are weighted at 5,

            The seasons two before and two after (1977 and 1981) are weighted at 4,

            The seasons three before and after (1976 and 1982) are weighted at 3, 

             The seasons four before and four after (1975 and 1983) are weighted at 2,

            And the seasons five before and after (1974 and 1984) are weighted at 1.

 

            Which makes the chart look like this:

 

 

 

 

 

 

 

 

 

WEIGHTED

1974

NL

1

4

1

0

4

5

4

5

1975

NL

4

3

1

0

7

4

14

8

1976

NL

4

0

1

0

7

1

21

3

1977

NL

2

4

1

0

5

5

20

20

1978

NL

2

4

1

0

5

5

25

25

1979

NL

4

3

1

0

7

4

70

40

1980

NL

4

2

1

0

7

3

35

15

1981

NL

4

2

1

0

7

3

28

12

1982

NL

4

3

1

0

7

4

21

12

1983

NL

1

4

0

1

2

7

4

14

1984

NL

1

4

1

0

4

5

4

5

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

246

159

 

            So the National League in 1979 scores at 246-159, a .607 percentage.  This is one of the highest points in NL history; it’s gotten above that by a few points a couple of times. 

            Summarizing the data, then. . . . .

 

            The National League has no competition in its first few years, but appears (by this one indicator) to be the stronger league in the 1880s, winning the ersatz World Series of the 1880s, against the American Association most of the time. 

            When the American League starts (1901), the NL is the stronger league (by this one indicator) from 1904 to 1909, with a peak score of .579 in 1907.

            In 1910 the American League begins to win the World Series almost every year, and the American League then shows as the stronger league every year from 1910 until 1952, with the exception of 1921 and 1922.   The American League was over .600, the National League under .400, in 1915, in every season from 1927 to 1939, and from 1947 to 1949.  The AL reached a peak score of .670 in 1932, dropping the NL to .330. 

            The NL became the stronger league (by this measure) in 1953, and remained the stronger league until 1982.   The NL is over .600 from 1963-1965, in 1976, and in 1979-1980, with a peak score of .623 in 1976. 

            After a three-year period in which the leagues appear to be nearly even (1983-1985), the American League moved back ahead (by this one measure) in 1986, and remained the stronger league until 2009.   The American League reached a peak dominance of .683, an all-time record, in 2004. 

            The National League had a little run, 2010-2012 (the San Francisco Giants years), scoring in the .540s all of those years.

            Since 2013, the American League has again been the dominant league, scoring over .600 in 2017-2018.  

            Thank you.  We’ll take a different look at the issue tomorrow.  

 
 

COMMENTS (7 Comments, most recent shown first)

shinsplint
Seems to me that sometimes the run differential between 2 teams in the World Series is often a better indicator of team strength than games won. Perhaps the most glaring example is the 1960 WS between the Pirates and Yankees where the Yankees outscored the Pirates 55-27, but lost.

And though the 1985 Royals won the World Series and outscored the Cardinals 28-13, The Cards could easily have won in game 6 but for an incorrect call.

There are other times where I suppose where the series run differential is not an indicator of a better team, such as when one team outscores another for a series due to piling up the runs on a day when the opposing pitcher(s) are just having one bad day, while in other games they score far fewer runs. So I'm not sure if run differential is on balance a better indicator than wins or not overall.

Also in the All-Star game, a 12-3 win could suggest a stronger league dominance as opposed to a 5-4 win. Or it could be an anomaly, but I'm guessing that a high run differential is more likely to be signal than noise considering that the high talent level of both teams should tend to minimize the game getting out of control.​
10:16 AM Jun 7th
 
tangotiger
Given Frank's results, I would say the "ballast" that Bill added of 1 W and 1 L should be more like 10-10 or 100-100. The #AspiringSaberist can calculate the actual ballast needed on his own and report the results to see what better matches the inter-league record.

I'm going to presume once Bill is all set and done, it's going to work out that way, that he'll do 5% World Series/ASG results and 10% other things and 23% other things, etc.

But still, I'd rather see the ballast applied at the component level (in this case anyway).
9:28 AM Jun 7th
 
FrankD
Interesting study .... looking at interleague play from 1997-2019 inclusive the league that won the most games vs. the other league won 14 All-Star games and 12 World Series (this is out of 23 seasons). If we limit the interleague difference to 10 or more wins then we have 19 seasons and the the league with the most wins won the All-Star game 11 times and won the World Series 9 times. At least over this time period neither winning the All-Star game or the World Series has little if any relationship to which league won the most interleague games (and therefore can be considered the 'best' league).​
12:38 AM Jun 7th
 
joeashp
I like to look at the rosters in the short term leagues, since players moved into those leagues and then back out. The 1884 UA was clearly poor. Not too many players moved to the NL and AA for 1885; Fred Dunlap was a very good player before and after 1884, but dominated the UA. The Players League looks to be the best league of 1890 - so many stars went to that league and remained stars afterward. Benny Kauff was the best guy in the Federal League, and was still very good in the NL but not nearly as dominant; not too many FL guys found success in the NL and AL. Also fun to see the AA guys like Browning, Orr, and King stay as top players in the PL.
10:26 PM Jun 6th
 
JohnPontoon
Something you (Bill) didn't say above, but that I think is worth noting, is that while the World Series has a single team representing each League each year, the All-Star Game has the very best players from ALL of their Leagues' teams on it. That factor alone, in my mind, would justify weighing the ASG more heavily than a WS game.
5:26 PM Jun 6th
 
evanecurb
Typo in my prior post: The last difference cited between leagues should say "[the NL had]...higher attendance."​
1:43 PM Jun 6th
 
evanecurb
There were differences between the AL and NL that were noticeable in the sixties and seventies. The differences were regularly mentioned on Game of the Week broadcasts in the seventies, and discussed occasionally in the Sporting News and Baseball Digest. The narrative went something like this: The NL of the early 1970s had a lower strike zone (attributed to the difference in chest protectors worn by home plate umpires), making it a "sinker/breaking ball league" while the AL, with its higher strike zone, was a "fastball" league. The NL had more scoring, more stolen bases, higher batting averages, newer ballparks, more artificial turf fields, and higher batting averages.
1:41 PM Jun 6th
 
 
© 2011 Be Jolly, Inc. All Rights Reserved.