We got to start thinking beyond our guns. These days are closing fast.
—"Lip" Pike Bishop in the "The Wild Bunch"

Here we are at the beginning of May with most teams having played about twenty-five games. So what have we learned? Anything besides Juan Rincon is a stupid, stupid man?

The Yankees seem to think that they've learned a great deal about their current team and decided to shake things up a bit in Billy Beane-like fashion.

Maybe the Yanks do have cause for some concern, but should the O's, D-Backs, Blue Jays, White Sox, and Nats feel that their early-season turnaround is anything more than Brian Roberts-esque, Tuffy Rhodesian, early-season mirage?

Does April mean anything to anyone besides May flowers?

I took a look at the team records for first twenty-five games of each season starting in 1901. I compared it to the team's eventual record to see what bearing it had. I also took the teams' expected winning percentage in the first 25 games and compared that to their overall record to see if maybe, just maybe, it presaged the teams' true capabilities better than just winning percentage.

What I found was the 25-game winning percentage correlates to the overall winning percentage somewhat but not all that well (coefficient of 0.625). However, it does ever so slightly better than expected winning percentage (0.617), which I did not expect.

Evidently, a large number of teams have changed their fates after 25 games and there was no indication, at least from their runs for and against, that they had it in them.

Take a look at the teams that improved the most after 25 games:

It's no surprise that the Miracle Braves are number one with a 432-point improvement after 25 games. It is a surprise that just four made the playoffs though and one was via the 1981 strike while another was a wildcard. I guess there are a lot of teams improving from abysmal in the small sample size to just plain bad for the season.

Anyway, one thing that is interesting is that the difference between the expected winning percentage and the actual 25 games into the season ("Exp Diff"), is positive for the teams with the greatest improvements.

Maybe this can be used to predict a post-25G improvement. Let's see…

Well, the results are that improvement after the first 25 games has even less to with a team underachieving their expected winning percentage through the first 25 games than our previous attempts at correlations through 25 games (coefficient of 0.470). So all of you using expected winning percentages in April to predict the division winners—you know who you are—can now stop.

Just for fun, I ran the same comparisons for the last 25 games of the season. Do good teams finish stronger or start stronger? They always same a win in April is worth two in the bush, or something like that.

I found that a team's record in the final 25 games correlates better to their overall record than their record in first 25 games (coefficient of 0.707). However, the expected winning percentage (0.696) still gets beaten out by the actual winning percentage.

Apparently, the final 25 games are a much better predictor of a team's overall success. It's too bad that the season's already in the books at that point though. It is odd given that you would think a lot of teams would be clinching early on in September and may rest regulars. Meanwhile others would know their fate is settled and may play less enthusiastically. And don't forget all the youngsters who are getting a try. And yet this time of year is more meaningful than April, odd.

Here are the teams with the greatest improvements in the last 25 games:

You'll notice that the 2001 A's make both lists, an odd feat. You may also note that six of these teams made the playoffs while only four from the previous list did. Not that that is conclusive, just interesting.