In 2004: 10 game stretch that was 2-8, both wins Santana.

In 2005, lost 2, Santana beats the Yankees, loose 6 more in a row.

Also in 2005 there were two other potential stretches, both involving Carlos Silva, one 10 game stretch were the Twins won only his two starts. And he started the 7th game that ended the 6 game slide mentioned earlier.

2006, no where near a 8 game loosing streak.

2007, a 10 game stretch where the Twins only won Boof’s 2 starts. 1 stretch were Santana’s win was in the middle of a potential 8 game loosing streak.

Make of that what you will.

No apology or retraction necessary. I think you’ve identified the disconnect correctly, and your math was never the problem. I read “.550″ and, this being primarily a baseball blog, saw that as a winning percentage, which we agree isn’t the greatest proxy for a team’s probability of winning their next game. Perhaps Joe intended something else. (Joe – You out there? Wanna chime in here?)

Saying that Rod Carew is the best Angel ever is like saying Pete Rose is the best Phillie ever. They were both greatly diminished from their best years by the time they joined the teams in question, and had both moved to first base, greatly reducing their defensive value. Also, both played in the same infield as a player who would be a much better choice as the team’s best player.

Of course, Pete Rose is not a Phillie – he’s a Red. Likewise, Carew is not an Angel; he’s a Twin. He’d be a reasonable choice as the best Twin/Senator position player ever; he’s not even close as an Angel.

I think we’re just having a miscommunication here. It seems to me that you’re the one who is linking winning percentage to probability. I (and it was my assumption that this was the case for Joe as well) am just referring to a hypothetical team that has a given probability of .550 of winning any individual game. It looks like winning percentage because, well, that’s how winning percentage is written, but I did not mean a team that has won 55% of their games.

I recognize that we can never know a team’s “true” winning percentage, but that’s not the point. Even if we flip a coin (to go back to that anology) 1,000 times and get heads 510 times, the coin’s “true” probability of landing heads is still .50 rather than .51 (discounting ridges and whatnot). I also recognize that assuming a team has a consistent .550 probability of winning a game discounts all types of factors quantifiable and otherwise, such as momentum, weather, home field, starting pitcher, injuries and so on. The point of my exercise, however, was to break it down to the basic mathematical principles in a quick and dirty fashion.

So no, actual winning percentage is not being considered (or at least, I’m not considering it). I do not consider a team that wins 16 of its first 20 games to be a “true” .800 team. Moreover, as you clearly also realize, we cannot ever know a team’s “true” probability of winning a given game. This is why I’m dealing in hypotheticals.

I think the root of our disagreement stems not from our interpretations of the math involved, but from our interpretations of Joe’s post. You seem to think that Joe is tying future winning probability to historical winning percentage, whereas I assumed otherwise. I think that you and I can both agree that to do so would be incorrect.

Unfortunately, only Joe can tell us what he actually meant, and I don’t get the sense that he reads this far down in the comments.

So perhaps I can, with my apologies, retract the part where I initiated my last comment with a reference to yours, and just merely say that, under the assumptions I’m using, the rest of the math (I think) holds true.

Jeez, it’s bad enough I have to deal with the dome, now I can’t avoid having people think my state is so small it’s merely a blip on the radar.

Just kidding. Thanks for the blog, Joe.