Parts I,II, and III.

Second Half Surge

Many in New York are worried about the Yankees' chances in the postseason given they way they have slumped in the second half. Meanwhile Red Sox fans are kvelling over their chances to win a much-coveted World Series crown given the way their team has played in the second half. Is there any basis in fact for these expectations?

Next, let's look at whether a team's improvement in the second half portends greater success in the playoffs. Here are the top ten second-half improvements for all playoff teams:

You'll note that only two of them won more than half of their playoff games.

Here are the ten worst:

You'll note that three teams, one more than on the top ten list, that won more than half of their playoff games. This doesn't bode well for using second half improvements as a predictor for postseason success.

Finally, I ran the numbers for second-half improvement and postseason winning percentage and the correlation coefficient was -0.073. If anything an improvement in the second half runs counter to postseason success ever so slightly.

Perhaps looking at the entire playoff population is misleading. Teams with huge leads at the break may tend to slump in the second half because of the large lead even though they have a superior team that shows up in the postseason. Let's limit the study to just those teams that were not in first place at the break but won the division/league in the second half (I'm sorry, I'm ignoring wild cards here—just too involved).

This at least reversed the correlation but it was still extremely weak (coefficient of 0.0076). So maybe we shouldn't keep blaming Billy Beane and the A's.

Pythagorean Pathos, Part Two (Say that three times fast)

Let's now compare how well second-half Pythagorean winning percentage predicts success in the next season as compared with a number of other factors.

Here are the correlation coefficients for each:

2nd-half Pythagorean PCT: .583
2nd-half winning PCT: .542
1st-half Pythagorean PCT: .538
1st-half winning PCT: .503
Overall winning PCT: .591

Pythagorean percentage correlates slightly better than winning percentage, but it loses out to the overall winning percentage (though none correlate very well).

To Be Continued…