Why the American League May Have Stronger Post Seasons

The answer may lie in the statistical analysis on the topic of how many games it would take each season to achieve a statistically significant result at the end of the year.

According to E. Ben-Naim and NW Hengartner, physicists at the Los Alamos National Laboratory , the best team will have the best record if the total number of games should be about the cube of the number of teams. So, for the National League the answer would be a 256 game season - sixteen teams - and for the American League - 14 teams - the answer would be a 196 game season.

Statistically, the larger the sample size - in this case, the more occurrences of an event - increases the probability of that sample size represents the truth of the matter. To illustrate that point: a player hitting 1.000 after the first game of the season is not indicative of where he'll finish because the sample size is too small (in this case it would likely be 3 or 4 at bats) or even of what kind of year he'll have a the plate. Statistical significance occurs, then, when the result is unlikely to have occurred by chance.

So, because the length of the season in both leagues is 162, the result at the end of the season is closer to statistical significance in the American League than in the National League, or the record of the best team in the American League is less likely to have been influenced by chance than that of the National League.

Because of that, the National League playoff representatives have records that are less close to statistically significant - or closer to having occurred by chance - than those of the American League.

The 2001 Arizona Diamondbacks won 92 games and won their division. The 2001 New York Yankees won 95 games and won their division. Which was the better team?

The divisional round of the baseball playoffs is a best three of five series, which would be shy of the number required to ensure the best of the two teams advances. This would seem to have been the case in 2001, when in the American League West there were two teams with more than 100 wins - The Seattle Mariners with 116 and the Oakland A's with 102 - yet both lost to the Yankees. When there is a small sample size in play, statistical anomalies can and do occur. After playing 162 regular season games, and 5 additional playoff games, the 116 win Mariners lost the American League Championship to the Yankees.

The Diamondbacks won 92 games and played the divisional round against a wild card team that had won 93. In this case, there was no statistical difference between their two records and as such, the teams would be considered even. It is not surprising then, that it took 5 games to determine Arizona as the winner to face the 88 win Braves.

The Braves took only 3 games to dispatch the Astros, even though the Astros had won 5 more games during the regular season. Since the National League schedule is 94 games shy of achieving a statistically significant result, that 5 game difference means little in terms of which team would be the better. The Diamondbacks took 5 games to dispatch the Braves in the National League Championship.

It took 7 games, but the Diamondbacks won the World Series over the Yankees. If the equation holds, if viewing the playoffs as a season in and of itself - and I would suggest it is - in theory, a statistically significant result could be achieved in an 8 game series. Since this is untenable, a 7 game series would be a decent compromise. So it would seem that the World Series winner - and the league championship series winner as well - would be close to being the best team between the two in the series. It seemed shocking at the time that Arizona would have won, but we can't know if the D'Backs should have been the underdogs.

The reason the 2001 World Series is looked at here is that while the American League is closer to having a statistically significant result at the end of their season, the Yankees - a team that had won 21 fewer games than the Mariners - emerged as the American League Champion, played a team that had won 3 fewer games and lost. Anything can happen - remember, statistical significance is not certainty, it just means that it's unlikely a result happens by chance.

In theory, the National League playoffs should be more volatile and result in a larger number of upsets than in the American League. Take a look at the 2003 Florida Marlins for example. They won 91 games, finishing as the National League Wild Card. The Atlanta Braves and the San Francisco Giants each had over 100 wins that season. The Central Division winner Chicago Cubs, won a mere 88 games, but dispatched the 101 game winning Braves in five games. The Marlins beat the 100 game winning Giants in four games. The Marlins then went on to win a 7 Game Championship series - made possible by the "Steve Bartman" interference call with the Cubs five outs from the World Series. This was as volatile a series and as improbable a match up as there could likely be.

The study is open to some criticism and while not the subject of this article, it should be discussed here. It makes a suggestion for efficient operations of the game, but the baseball season is based on the business end of things: it may not be an efficient model, but it is a legitimate business model, without which we would not have this discussion. Baseball is, after all, a business and the study's suggestion of sequential elimination doesn't suit the business' need for largest geographic appeal of their product nor the actual performance of the games.

Further, baseball teams no longer play a fixed number of games against the same rivals the same number of times. This was the case prior to the creation of the three divisions in baseball, however in today's game there is the "unbalanced schedule" between divisional opponents and since 1997 inter league play has further eroded the statistical basis of standings. However, baseball is better positioned statistically to compare "best team" scenarios than other games because of the large number of games played - remember, the smaller the sample size, the less true the result and the more statistical anomalies can occur.

The truth is there, though, that the National League standings have a greater degree of chance built into them than do the American League standings, meaning that it leaves the National League Champions up to a greater degree of chance than the American League Champions. Which increases the likelihood that a lesser team becomes the National League Champion.

REFERENCES AND RESOURCES:

E. Ben-Naim and N.W. Hengartner, Phys. Rev. E 76, 026106 (2007) "The Efficiency of Competitions,"
URL: http://works.bepress.com/cgi/viewcontent.cgi?article=1012&context=ebn

Physorg.com, "Better Baseball, Choosing the Champs"
URL: http://www.physorg.com/news105010579.html

Statistical Significance,
URL: http://en.wikipedia.org/wiki/Statistical_significance