DLD 6.18.08 remember that one time

If that’s a veteran player and it’s 9-1, that’s a double.

Way to buck the trend, Gonzo.

he’s certainly earned it over buck and suzuki.
they should squeeze duke in this year, anyone who wants a dick williams can just dye the hair on a carney bobblehead.

That’s what happened to me the last time I came back from a foreign country with the runs.

I guess it’s safe to say Joe Garagiola wasn’t in the booth last night. At least they didn’t guess that Kurt Suzuki is Italian.

of their games on MLB.tv (unlike the A’s announcers) so you get to hear a lot of them if you follow the Dbacks.

I think is marginally less Italian sounding. But yeah, Duchscherer has to be a close runner up.

(Yes, definitely German.)

Now that I actually think about it seriously, the least Italian-sounding last name I can think of is “Worth”.

kept updating the plane on the status of the game. When we got in the cab to head home, the cabbie couldn’t stop talking about the damn Celtics. Bah.

i’m not a golf guy but this US open was great entertainment.

What a remarkable athlete.

welcome back!!!! – meant to say that!!!

who are you again?

...where’s my present?

but haven’t people been looking over you for long enough that you’re accustomed to it by now?

I just tried to click on this link and the evil corporate web police gave me a “you can’t go here message”.

If AN gets blocked on the office computer I may have to quit. Although, my productivity would probably go up so much that I would get a promotion and could afford to take more time on home computer, so maybe it would be a wash.

is a reference to a running joke on SNL where Chris Farley played himself giving Ray Fosse-like interviews (“Remember that one time when…that was awesome.”
Also, it’s Sir Paul’s birthday today.

I hope they had a redundancy in place in the interim

would you like a Richie Sexson for a present? I hear he’s free. He can change lightbulbs without standing on a chair.

since after the oceans rise due to climate change, we’ll all be in the Central time zone.

Where do I send the invoice—to Blez?

13% for the game vs. 11% for the season. Maybe he’s finally turning it around.

Almost.

Every time I start to feel sorry for him, I think of $18,000,000 reasons why I shouldn’t feel sorry for him.

Okay, so when he retires/traded/DFAed, he’ll never be able to set foot in San Francisco again. That’s not the end of the world. Besides, LA loves him.

But I’m sure Barry could have struck out swinging at a wild pitch just as effectively as Horowitz.

Except in our case, the whole school knew. It was a planned thing where certain students would “die” and walk out of class in dramatic fashion with the grim reaper. Then someone came in and read an obituary for them. There was a big assembly outside where they had brought in a wrecked car and the dead students were strewn around it. A helicopter landed on our field and took away one of the victims. It was a little over the top, to say the least.

I wasn’t quite cynical enough to play hackeysack behind the bleachers while this went on, but I did bring a book to read.

“And that’s why you always leave a note!”

when it was reported last month. Lying and manipulating children are means which are never justified by the ends sought—ends which, as Jon points out, are very unlikely to be achieved. And the whole thing was promulgated by their guidance counselor. That part shocked me at first, until I remembered that my High school guidance counselor (me and the rest of the kids last named A-G) was the school football coach.

This should work out well.

Each league is a collection of teams and each team has a true talent level (that we’ll assume is constant, at least for the span of interleague play). We want to compare the average true talent level of AL teams to that of NL teams.

Looking at the results of baseball games is a way to measure the true talent of teams, but it is noisy. As you pointed out, there is some inherent amount of randomness in any game between two teams. Starting with a small number of games, you have a binomial distribution, but it converges to a nice normal distribution (yay central limit theorem!) when the number of games gets large. If two teams played each other 115 times with a final record of 67-48, you could reasonably say (using somewhat sloppy statistics) that the better team had a win probability for each game of 0.582 +- 0.046. So you have 1.8 sigma confidence (about 96%) that one team is really better than the other.

However, this isn’t a case of one team beating the other 67 times out of 115. Getting back to the idea of each league containing many teams with different true talent, you need to add some additional error due to the randomness of the matchups. If the talent distribution for teams in each league could be described by a normal distribution as well, then one could probably come up with some easy way to add the variances, but I imagine that the real distribution of talent among teams is not so well behaved.

However, in the end every team plays the same number of interleague games, so I don’t think the extra error biases the result. There is still some extra error, so your confidence level isn’t 96% anymore, but I think that the interleague results to date still make a strong argument for the AL being better than the NL.

I think colin pretty much covered the basic ideas.
Math part:
On your middle question (“Presumably tennis player A beating tennis player B 58% of the time is much more reliable than baseball team A beating baseball team B 58% of the time”), No, not really. The important things about any binary statistic (heads or tails, hit or out, win or loss are):

1. A standard deviation for one trial is sqrt(x(1-x)), where x is the probability of success. Unless x is near one of the extremes, this is always pretty close to .5. For example, when x = .58, you get .49. (The intuition here is: in one game, the optimal estimate for how many wins team A will get is (around) .5. But since it’s impossible to get half a win, you will on average (in fact, always) be off by about half a win.)

2. For larger numbers of samples, one standard deviation in absolute terms increases in proportion to the number of samples. Equivalently, in percentage terms, it decreases in proportion to the number of samples. So after 115 trials, one standard deviation is around .5*sqrt(115) = 5.4 wins, or .5/sqrt(115) = .047 in win percentage.

Once you understand this, you might wonder how much to regress the sample estimate (58%) to the mean. There’s no clear answer to that; it depends on what your estimate of the relative strengths was prior to those 115 games.

Math/baseball part:
If Team A beats Team B 58% of the time (over a very large sample size), how much better is Team A than Team B? Bill James looked at this (or the equivalent question), and came up with one formula that fits the data pretty well. Using the simplified form, the difference in head-to-head winning percentage is about twice the difference in expected winning percentage over a season, i.e. a .540 team will beat a .460 team about 58% of the time head-to-head.

Baseball part
Going back to the actual case, where we’re looking at leagues, rather than individual teams: in addition to the sampling issue colin mentions (teams have played different numbers of interleague games at this point), you should probably also consider where the games have been played. In addition to the usual home field advantage that all teams seem to have, in interleague play there is an additional effect because of the different roster construction between the two leagues: In NL parks AL teams are forced to sit one of their presumably best hitters and have their pitchers bat, while in AL parks NL teams are putting someone in the lineup who would normally be on the bench. (I tend to think that on the whole this favors the AL team.)