because Sizemore has the higher ceiling.

Uh oh. Now you've done it. Now you made them angry.

"Uh oh. Now you've done it. Now you made them angry."

Ha, tell me about it. I was stunned to come back to my computer and not see 50 new posts ripping me.

Reyes' batting line since he "broke out" last June against Arizona. This covers his last 590 plate appearances.

.338/.393/.542.

Listen, Grady is a fantastic player and I wouldn't say that someone who choose Grady over Reyes was making a mistake. You really couldn't go wrong with either player. That said, anyone who thinks Sizemore is clearly better than Reyes is wrong in my opinion. What I mean, choosing Sizemore over Reyes isn't a no-brainer and anyone who is arguing it is is wrong.

In other words, no matter how big the sample seems, this is still a very small percentage of total games played. So is it like calculating the odds of a coin being flipped heads or tails 575 times out of a thousand? I honestly don't know.

If .575 vs. .425 over 3 times the length of an individual season is too small a sample size to be meaningful, then no single season record is sufficient to be meaningful to distinguish the quality of any two teams. Do you not believe that the Yankees and Twins last year were clearly better than the Royals and Devil Rays?

Sizemore has the higher ceiling.

That's mathematically impossible.

Reyes's ceiling is Grady Sizemore, but with more speed.

Do you not believe that the Yankees and Twins last year were clearly better than the Royals and Devil Rays?

Yes. I do. But not because of how they played against each other. The Devil Rays or Royals could have beaten the Yanks or Twins in the season series and I wouldn't care because there was more datat to fall back on. The problem with the interleague play, I think, is that it's analogous to how teams did in a season series. It's about 10 percent of games played, I believe. So it doesn't matter how big you make it by folding in all of major league baseball. You still have to accept that you can draw a definitive conclusion from that amount of games played.

The Mariners were 14-4 in interleague play last year. Can we conclude that this means they were better than any NL team? If not, what's different from that and what were doing with the interleague games from all teams? IOW, it's still the same (small) percentage of games, it's just more teams.

Basically, five teams were responsible for all the differential between the leagues last year. And these teams weren't even the best in the AL: Red Sox, Twins, Tigers, White Sox, Mariners.

Yes. I do. But not because of how they played against each other.

You missed Dan's point - the sample in question isn't the season series - it's more than three times the length of the season. The season itself - not the season series between two teams.

If .575 vs. .425 over 3 times the length of an individual season

Put another way, and I'm asking here mostly, not trying to be snarky, what's the difference between doing this and looking at how the White Sox are hitting now, for example, and saying they really are a .220-hitting team because the sample right now is X-times the length of an individual player's season?

Put another way, and I'm asking here mostly, not trying to be snarky, what's the difference between doing this and looking at how the White Sox are hitting now, for example, and saying they really are a .220-hitting team because the sample right now is X-times the length of an individual player's season?

1. The sample of an individual players' season is much less than the sample of a team's entire season.

2. The sample in question was three times larger than a team's full season; the White Sox are currently at 1.5 times a single player's season.

Most importantly - there are peripheral indicators, which is the real reason we think there's a league differential. The interleague play results are glaring and indicative, but they're not the basis of conclusion.

As an added plus, McCann is a quiet, humble kid on a team full of big egos

I never imagined the Braves as the big-ego team. Thanks Nate Silver, that'll help me hate them more.

In other words, no matter how big the sample seems, this is still a very small percentage of total games played. So is it like calculating the odds of a coin being flipped heads or tails 575 times out of a thousand? I honestly don't know.

If .575 vs. .425 over 3 times the length of an individual season is too small a sample size to be meaningful, then no single season record is sufficient to be meaningful to distinguish the quality of any two teams. Do you not believe that the Yankees and Twins last year were clearly better than the Royals and Devil Rays?

There are several reasons why Dan's analogy is flawed, insofar as using it as sufficient sample size to argue about an entire league's quality. Will come back later tonight to post.

I never imagined the Braves as the big-ego team.

My fellow Brave fans and I were also wondering about that line. Sure, some of the Braves are a little cocky, but I've never gotten the impression they were any more big-egoed than most teams.

If .575 vs. .425 over 3 times the length of an individual season is too small a sample size to be meaningful, then no single season record is sufficient to be meaningful to distinguish the quality of any two teams. Do you not believe that the Yankees and Twins last year were clearly better than the Royals and Devil Rays?

There are several reasons why Dan's analogy is flawed, insofar as using it as sufficient sample size to argue about an entire league's quality. Will come back later tonight to post.

I'm back. That's not how one determines sufficient sample size. You need to determine first what question you are asking, and then what conclusion you are reaching.

In your example, using the Yankees and the Twins' regular season record to determine their superiority over the Royals and the D-Rays is reasonable in so far as one believes that you can generate a reproducibility with the results. One factor in that favor is that it is specifically the Yankees who are playing 162 games and that they are playing a fairly similar schedule to the D-Rays. In that manner, we're making the assumption that the parameters are reasonably fixed enough with the key variable being the team (i.e. Yankees play a schedule 162 times, then D-Rays doing the same thing 162 exactly the same way.) Of course, there are lots of hidden assumptions behind such a comparison.

All these assumptions are completely lost when you use interleague play schedule.
First, the individual members are heterogenous (each league composed of a number of teams of varying qualities).

Second, you are not sampling over all of matchup space (i.e, not all the possible permutations are tested, in fact, it's only a small fraction of permutations that are tested).

Third, you have a major migration effect when you are sampling through 3 seasons (players come and go, new prospects are broken in, old players retire, the sets which you are studying are not closed). One only needs to see how different this season has played out from last season to know how much a league can change from year to year.

Finally, they're tested against each other, without a fixed control to test against. This in itself introduces a difficulty in the analysis depending on what conclusion you're trying to reach.

The trouble with all of this makes generating translations and adjustment parameters extremely tricky. Certainly it is not as simple as the analogy of "3 times the length of an individual season is too small a sample size to be meaningful," which is the wrong way to frame this.

BTW, since this is sort of like a Mets thread:

Carlos Gomez tonight 3-4, 1HR (apparently was a bomb).

Francisco Pena, 2-4, 1st HR. Congrats.

I am guessing now - I would have to look it up - but after around 5 years I think that the chance that a well-above average pitcher is even still pitching at all is around 50%.

Here are the AL pitchers who pitched at least 100 innings in 2001 with an ERA at least one run below league average:

Mark Buehrle
Roger Clemens
Kelvim Escobar
Freddy Garcia
Roy Halladay
Tim Hudson
Pedro Martinez
Jamie Moyer
Mark Mulder
Mike Mussina

And these teams weren't even the best in the AL: Red Sox, Twins, Tigers, White Sox, Mariners.

Am I missing something? Three of those teams won at least 90 games last year and another was a contender to the last couple weeks of the season. The only 'great' teams of the AL you're missing (in 2006) is the Yanks and A's, and off the top of my head, I don't recall how they did in interleague.

Was that the worst trade of the past 25 years?

November 14, 2003: A.J. Pierzynski traded by the Minnesota Twins with cash to the San Francisco Giants for Joe Nathan, Francisco Liriano, and Boof Bonser.

This one was worse. At least Colon has been a useful pitcher. Plus it was the Expos.