THE BOOK--Playing The Percentages In Baseball

Oops!  I counted them as if they were home in Montreal.

But Wikipedia says that Puerto Rico doesn’t go on Daylight Savings Time, which means that they’d be on the equivalent of Eastern Time during the season.  If I’ve got that right.

In any case, I did treat them as home games for the Expos.

Well, I think I would treat them as home, since they were of long-enough consecutive duration.  It’s not like the P.R. games where spread across the season.  If you play, say, 9 straight games in the same park against 3 different opponents, that’s “home”, or its equivalent.

Anyway, that’s probably your source of discrepancy.  The others might be the Tokyo games, and those would be road for both (as they are really neutral-site games).

Yup, I agree that they should be treated as home games.  I’m not sure if Dr. Winter did, also ... I’ll ask him when he gets back to me with the abstract.  If he did, then that might indeed be the discrepancy.

I counted the Tokyo nominal home teams as “real” home teams, too, but there’s no “circadian advantage” in those games, because both teams had a chance to acclimatize, so that won’t affect the results at all.

Thanks for the catch!

I remember reading a newspaper article last season about this issue regarding the B.C. Lions of the C.F.L. The team was accused of trying to schedule their home game against an Eastern division opponent as late as possible to take advantage of time zone lag. They’d start some games at 07:30 P.M.( really later once you get all the anthems out of the way) much to the consternation of the visiting team.

Basically by the time the game finished it would be about 02:00 A.M. Eastern time for the teams who had just flown in the previous morning. 

I’m not sure how you feel that late at night, but I’m certainly not up for a football game.

I’m not sure baseball is the best sport to study circadian advantage given the teams stay in one time zone for 5-6 days. Also isn’t it more difficult to go West verses East?

If the data indicate “no effect” and the author says that there was no effect (see the comments section on Phil’s blog), why is it being reported that there is an effect?

And what does this, also in a comment by the author (of the study), mean?

It’s the 2 and especially 3 hour games that tend to show changes.

Does he mean the duration of the games? Isn’t that pretty much all the games?  The fastest games these days (the period of the study) are 2 hour.  3 hour games are at the upper end, no?  IOW, a 2 hour game is a super fast game and a 3 hour game is pretty slow (not the slowest).  So how does he group 2 and 3 hour games in that sentence, and what does he mean?

I am pretty sure I did a similar study a long time ago and came up with no effect.

Plus I would think that there are all kinds of confounding effects that you would want to look at whether you did or did not find an effect, in addition to controlling for the quality of the teams (which, I agree with Phil that they should not be a problem in a large study, but nevertheless I would want to make sure, since that is certainly a potential for bias in the study).  For example, was the travel the night before the same day of the game?  How far was the travel?  For example, TOR and MIA are in the same time zone, but it is a long trip.  How long is a team on the road or at home.  All of these things (and more) could potentially affect fatigue and therefore winning percentage.

Also, what is the reason for a “circadian disadvantage?” That is important.  For example, it may be that “regular people” who go to bed a certain time every day and wake up at a certain time for work are affected when flying across time zones, but baseball players who can wake up at 12 or 2 in the afternoon and go to bed whenever they want are not.  And what if I fly to Arizona from California which is technically in a different time zone but is on the same time (because ARI does not use DST)?  Would that affect me?  IOW, is the effect just from flying across the country and having the sun set and rise at a different “time” than I am used to, or is it from the fact that the time of the games (and everything else, like breakfast, lunch, dinner, etc.) is different than I am used to?

And, BTW, the home team has a slightly bigger advantage on the first day of a series.  That extra advantage diminishes as the series goes on.  That is presumably because the road team gets more used to the park each day, or maybe they are a little less tired.

mgl, he means the games with the 2- and 3-hour time zone changes, where one team has at least a 2 hour circadian advantage.

I’m not sure where the reported effect actually is.  The numbers quoted in the press release appear to show an advantage, but don’t (once you consider home field).  And the 3-hour numbers don’t show any effect either.

Oh, OK, that is what I thought at first, but the wording was kind of weird.

So, when he presented the paper, did he say, “I could not find any statistically significant effect on a team’s WP as a function of how many time zones they recently traveled?”

If the 3 hour numbers don’t show any effect (is that according to your numbers Phil, or his, although they should be the same of course), why did he specifically say that they did?

How can a study be reported as showing one thing and the data and the author show another?  I don’t understand this at all. Have we all lost our minds?

I think this is what happened:

Dr. Winter wrote the study, which is unavailable.  However, a press release split the numbers into home and road.  Breaking those numbers down (my first post) showed no “circadian advantage,” that is was all home field.

I tried to duplicate Dr. Winter’s numbers (my second blog post), and appeared to be off by 30 games home and -30 games road (the total was almost exactly right).  I wondered if I made a programming error.  Dr. Winter wrote that my numbers were right (implying that the press release got a digit wrong in its reporting).

However, Dr. Winter says that although I didn’t find an effect, that’s because I calculated an effect for ANY jet-lag advantage, even one time zone.  He said that most of the effect is for 2 or 3 hours.

Today I put up a third post, where I didn’t find any effect for 3 hours either.  So I’m not sure where the significant finding comes from.

Perhaps Dr. Winter will comment again ...