Is the American League’s dominance coming to an end?

by KerryWhisnant

Even though the National League has claimed a World Series title in five of the last ten years, the American League has generally been recognized as the better of the two leagues during the last decade or so. The first evidence people point to is the streak of thirteen straight All-Star games without a loss – finally broken last year — but even more compelling is the AL’s 1345-1175 inter-league record over the last ten years, a winning percentage of .534. There is only one chance in about 2600 that the leagues are in fact equal and this is due simply to random chance. But does the All-Star win by the NL indicate the senior circuit has finally caught up?

Of course the results of a single game do not make a trend, especially in baseball. And last year the AL still had a healthy 134-118 record in inter-league games. But an analysis of Pythagorean expectation suggests that the NL may in fact achieve parity with, or even a modest superiority to, the AL in 2011.

It is well documented that Pythagorean records are a better predictor of a team’s record the following year than their actual record. What does a Pythagorean analysis say about the relative strengths of the two leagues? The following table shows the actual inter-league record of the American League (in terms of games above .500) and the Pythagorean expectation for each year since inter-league play began. (The calculation of the Pythagorean expectation is not straightforward, since the Pythagorean expectation records average about 82 wins per team per season, instead of the 81 wins you would expect. The Pythagorean records have been normalized to an average of 81 wins per team.)

Year Actual Pythag.
1997 -20 - 7
1998 4 - 7
1999 -19 -13
2000 21 7
2001 12 0
2002 -  6 3
2003 -22 - 6
2004 2 4
2005 20 37
2006 56 43
2007 22 32
2008 46 47
2009 24 24
2010 16 2

The Pythagorean expectation for the inter-league record tracks well with the actual record – in fact, the correlation between the two is 0.87. But it is the differences that are important. The following table compares  (i) the difference between the Pythagorean inter-league record and the actual AL inter-league record for the previous year with (ii) the difference between the AL inter-league record for that year with their inter-league record for the previous year.

Year Pythag. minus actual

for previous year

Current year minus

previous year

1998 13 24
1999 -11 -23
2000 6 40
2001 -14 -  9
2002 -12 -18
2003 9 -16
2004 16 24
2005 2 18
2006 17 36
2007 -13 -34
2008 10 24
2009 1 -22
2010 0 -  8
2011 -14 ?

The trend here is that whenever the Pythagorean inter-league record is better than the actual record in a given year, the following year’s inter-league record tends to move in the direction of the Pythagorean record. This can be seen by noting that whenever the number in the first column is significantly positive, the number in the second column is very likely to also be positive, and similarly for negative values. In fact, for the ten years in which the Pythagorean inter-league record differed by more than two games from the actual inter-league record, the following year’s inter-league record moved in the direction of the Pythagorean record nine times.

Using a linear regression on the data in the above table, we can make a prediction for the 2011 AL inter-league record using last year’s Pythagorean record — filling in the question mark in the table above. The result: the Pythagorean prediction is for the AL to have 12 fewer inter-league wins in 2011 than in 2010, i.e., a 122-130 inter-league record. So this Pythagorean projection suggests that the NL may finally turn the tables on the AL in 2011.

Of course the data is sparse since inter-league play began just 14 years ago, and any predictions are not expected to be exact. I have not done a precise mathematical calculation, but from the spread of the data, I estimate that the odds of the AL winning fewer inter-league games than last year (i.e., having a record worse than 134-118) is about 92%, and the odds that the NL will win more inter-league games than the AL this year is about 73%.

So, will the NL be able to compete with the AL this year in inter-league play? Perhaps surprisingly, the odds appear to be in their favor.

Recent Posts

4 Responses to “Is the American League’s dominance coming to an end?”

  1. Hartvig Says:

    I’ve wondered about this for a few years myself, especially with the seeming outflow of the AL’s most dominant pitchers to the NL (Johnson then Martinez then Santana then Halladay then Lee- sorry, couldn’t resist) coupled with the rapid development of some of the young NL pitching like Lincecum and what seemed for a while the stalled or slower improvements that were being shown by guys like Greinke & Hernandez. The general rookie talent pool in the NL has seemed deeper (at least at the high end) in recent years as well:

    in 2006 the top 7 ROY vote getters in the NL were: Ramirez, Zimmerman, Uggla, Johnson, Ethier, Cain & Fielder. The AL had a decent class too: Verlander, Papeldon, Liriano, Johhima, Weaver, Markakis & Kinsler. In 2007 the AL put Dustin Pedroia & Delmon Young up against Ryan Braun & Troy Tulowitzki. In 2008 you had Evan Longoria, Alexei Ramirez & Jacoby Ellsbury against Joey Votto, Jair Jurrjens & Jay Bruce. In 2009 Andrew Bailey, Elvis Andrus & Brett Anderson vs Tommy Hanson, Andrew McCutchen & Colby Rasmus. Last year was Neftali Feliz & Austin Jackson vs Buster Posey, Jason Heyward, Mike Stanton & Madison Bumgarner. I have to score that a win (or, at worst, maybe a tie in 08) for the NL in every season. It would seem the tied has to be changing sometime soon.

  2. KerryWhisnant Says:

    Hartvig,

    I decided to quantify your observation. Using WAR from baseball-reference.com (I know this violates Chuck’s new guidelines for the use of WAR; sorry Chuck :-) ), I took all players in 2010 with WAR 2.0 or greater and tabulated how many had changed leagues in the last three years (including the recent off-season) or entered the league in the last four (since it usually takes them a season or two to get going).

    The net result: the NL had a net gain of about 24 WAR, which is equivalent to 48 in games above .500 (since each NL win subtracts an NL loss). This is roughly consistent with the change in games above .500 for the AL of close to fifty down to the teens, and suggests it will be close to zero this year. Of course the many players with WAR less than 2.0 might change the result, but it accounts for the impact players.

  3. Hartvig Says:

    Kerry- And I would guess when the class of ‘10 gets a year or 2 under it’s belt then impact will be much larger. I think you may actually be understating the swing. I would bet than the NL will win more games than the AL this year. It’s possible that Boston’s & New York’s payrolls plus the possibility of a resurgent Kansas City might offset some of the National Leagues gains over the next few years but I’d feel pretty comfortable putting money on the NL for the next 5 to 7 years unless something fairly big happens.

    Geez, and I even missed CarGo in the comparisons, I see

    And I know: it’s tide, not tied

  4. JohnBowen Says:

    Donwert wrote:

    “The problem with comparing NL and AL rookie crops is that their records are largely made vs. their respective league opponents. If the NL is weaker, NL rookies might well have better numbers than their AL counterparts. But this would not necessarily mean they are better players, head-to-head. I note that, as of the conclusion of play on June 21st, the AL leads the NL 56-38—a winning percentage of .595. No doubt the cycle will one day turn around—that’s inevitable. But it doesn’t look like it will happen this year.”

    Welcome to DC, Donwert! If you comment on any of mine articles, I can approve you and then you’ll be good to post.

Leave a Reply

Advertisement