Do Champions Choke?

As of 1990, there had been 24 Super Bowls and only 4 champions were able to repeat their feat the following year. In baseball, of the last 26 World Series champions (1964-1990), only 4 have repeated. No professional basketball team repeated as NBA champion between 1969 (the Boston Celtics) and 1988 (the Los Angeles Lakers). This doesn’t mean that champions choke or that the teams in each of these sports are becoming increasingly equal in ability. Instead, these can be explained as regression toward the mean. There is a considerable amount of luck involved in winning a championship. Injuries, bad bounces, and questionable officiating all play a role. The team that ends up as champion is more likely to have had its share of good fortune. The next year, some other team will probably get the breaks.

There is plenty of other more detailed evidence of regression toward the mean in sports. Of those major league baseball teams that finish a season with winning records, two-thirds win fewer games the next season. Of those teams with losing records, two-thirds do better the next season. Of those teams that win more than 100 out of 162 games in a season, 90 percent will find their record deteriorating next season. Looking at individuals, there were 28 major league baseball players who batted above .300 in 1979, and 23 of those 28 had lower batting averages in 1980 than in 1979. This is to be expected because someone who bats .320 in any given year is more likely to be a .300 hitter having a good year than a .340 hitter having a bad year, just because there are so few (if any) .340 hitters around. The regression-toward-the-mean fallacy is to conclude that good hitters’ skills deteriorate. The correct conclusion is that those with the highest batting averages in any particular year aren’t really as skillful as their high batting averages suggest, in that they undoubtedly had more than their share of good luck that year.

Smith, Gary. Statistical Reasoning. 3rd edition. United States: McGraw-Hill, Inc., 1994.