Comments on: Color By Numbers: There’s No Place Like Home

By: RIYank

RIYank — Thu, 20 Oct 2011 20:54:17 +0000

Oh, and by the same token, the fact that the overall winning percentage of home teams in the Series is so close to overall winning percentage of home teams in the regular season makes the Series trend much stronger even than the chi square test indicates.

By: RIYank

RIYank — Thu, 20 Oct 2011 20:52:44 +0000

[10] Yeah, well. The test is a little more complicated and I don’t feel like doing it, but to me that looks like a bell curve around .55, with three of the four lowish ones all clumped together, just coincidence.
If there were some antecedent reason to expect the middle games to be different, then I’d be more inclined to think it was ‘real’.

By: William J.

William J. — Thu, 20 Oct 2011 20:00:46 +0000

[9] What he said!

Seriously, I’d be inclined to accept the middle games being the product of chance, if 1,2, and 6 didn’t all have a much higher winning percentage than 3,4, and 5.

By: RIYank

RIYank — Thu, 20 Oct 2011 17:56:32 +0000

My bet is that the loss of the home field advantage in the middle games of the Series is: pure chance.

I do think the total number of WS games played is large enough to support the conclusion that the home field advantage overall is a real one. Here’s one way to quantify this judgment.

The null hypothesis is that there is no home field advantage at all, that knowing which team is the home team in some future game will give you no information about who will win. Our alternative is that there is a home field advantage (it does not include that there is a disadvantage — it’s obviously reasonable to use a ‘single-winged’ test). The data in William’s table support rejection of the null hypothesis with significance at the .005 level. That means if you just modeled the games by a coin flip, the chance of getting a result as skewed toward the home teams as the historical results are is tiny — a half of a percent. It’s extremely unlikely that the results are just due to noise.

(The confidence interval here is .513 to .591. That is to say, we can be 95% confident that the real advantage to the home team is somewhere in there — the best guess is that it’s the 55.2% that we actually saw, but it wouldn’t be hugely surprising if the true advantage were as small as a little over a percent or as large as 9 percent.)

But the sample size for the middle games doesn’t support any interesting conclusions about those games. If we just take the simple hypothesis that in every single game being the home team gives you a 55% chance of winning, we should not be at all surprised to see that in a historical data set there are some spurious patterns, patterns like the loss of the advantage in middle games. My bet is that’s what’s going on in the data.

By: William J.

William J. — Thu, 20 Oct 2011 17:32:33 +0000

[6] Because schedules are so different, I don’t see the point in assigning home field based on an apples-to-oranges comparison. For example, winning 94 games in the NL Central is not better than 91 games in the AL East, so why grant home field in that manner.

I think there are two legit approaches: (1) develop a saber-type formula to normalize differences and come out with a rating; or (2) have some fun with it and attach to the ASG, which does have way of representing which league is better.

By: ms october

ms october — Thu, 20 Oct 2011 17:01:55 +0000

[6] ugh – sights = sites

By: ms october

ms october — Thu, 20 Oct 2011 17:01:23 +0000

[5] perhaps when the al was winning all those asgs :}
i am definitely not a fan of the random approach.
but there is really no good reason why home field cannot go to the actual team with the best record.
the nba has more teams in the playoffs and thus more potential sights for where the finals ends up being so the logistics argument is crap to me.

the cards got home field this year because their current biggest rival’s chubby 1b hit a home run in the asg – i don’t see how the cards had anything to do with that :}

By: William J.

William J. — Thu, 20 Oct 2011 16:16:15 +0000

[3] I am actually a big supporter of the ASG link because I do think it spices the game up. I also think the ASG outcome can indirectly funnel HF to the better team, and is at least much better than the random approach.

By: William J.

William J. — Thu, 20 Oct 2011 16:13:42 +0000

[1] In the link, I address that point in more detail, but I don’t think NL and AL schedules are congruous enough to make the best record approach more meaningful than random. Also worth noting that the ASG method has given home field to better record in 7 of 9 years.

[2] Perhaps, but the uptick really isn’t that big. Having larger crowds could explain the difference (although that could also incorporate umpire bias). It would be nice to do a study comparing HF based on attendance figures, but don’t know of a database that has such information.

By: ms october

ms october — Thu, 20 Oct 2011 16:09:25 +0000

interesting piece and analysis william.

i have never liked the idea that the asg is tied to homefield and have become even more opposed to it over time.
a team with a slightly better than middling record like the cards should not have homefield advantage.

[2] that seems to make a lot of sense.

By: Jon DeRosa

Jon DeRosa — Thu, 20 Oct 2011 15:45:23 +0000

linking this back to the study that showed that almost all of the home field advantage is in biased umpiring/officiating, i wonder if the heightened atmosphere of the world series, the extra amped up crowds, influences the umps even more than usual, accounting for the uptick in the WS home field advantage over the regular season.

By: Alex Belth

Alex Belth — Thu, 20 Oct 2011 15:17:15 +0000

Nice piece, man. I know the Lords of the Realm want the All Star Game to mean something so they can draw ratings but it seems to me that the fairest way to determine Home Field is that it should go to WS team with the best regular season record.