This falls under the category of just shaking the numbers and seeing what falls out. It has very little analytical value. While I'm keeping track of attendance, I'm also keeping track of lots of other things, so I thought I'd check average temperatures and weather in all MLS regular-season games for 2003. Yes, I have too much time on my hands. Sue me. Anyway, here's the monthly average temperature at gametime (all according to the official boxscores, no heat or wind chill indexes, conditions may have changed during the game, it may have felt hotter or colder to you or to those on the field, whatever): Month......Avg. April......64.4 May........68.0 June.......76.3 July.......80.1 August.....84.0 September..72.6 October....65.2 TOTAL......73.5 From this, I conclude that America warms up as the summer goes along, with the highest temperatures reached in August. Then, it gets cooler. As for which team had the highest average temperature at kickoff of its games in 2003: Team...........Avg. Dallas.........85.1 Kansas City....75.6 San Jose.......75.6 Colorado.......75.3 Los Angeles....74.8 Chicago........72.5 DC United......71.9 New England....69.9 MetroStars.....68.6 Columbus.......66.2 TOTAL..........73.5 From this I conclude that it gets hot in Dallas in the summer. Here are the 10 hottest games of 2003: SAT...08/23/2003...Columbus at Dallas..........8:00..104 Clear SAT...07/19/2003...MetroStars at Dallas........8:00...99 Partly Sunny FRI...08/22/2003...Colorado at Kansas City.....7:00...97 Sunny SAT...08/16/2003...Kansas City at Dallas.......8:00...96 Clear SAT...06/28/2003...Kansas City at San Jose.....7:00...93 Sunny SAT...07/05/2003...MetroStars at DC United.....4:00...93 Sunny SAT...08/16/2003...San Jose at New England.....4:00...92 Sunny SAT...05/31/2003...DC United at Dallas.........8:00...91 Sunny SAT...06/21/2003...Colorado at Dallas..........8:00...91 Clear FRI...08/08/2003...San Jose at Kansas City.....7:00...91 Sunny Incidentally, the "official" gametime temperature at the "Heat Stroke" game at Giants Stadium at Noon on Wednesday, August 13 was 83. And the 10 coldest games: SAT...04/05/2003...Los Angeles at Columbus.....4:00...41 Cloudy SUN...10/26/2003...Chicago at Columbus.........4:00...49 Cloudy SAT...04/19/2003...Columbus at New England.....4:00...50 Sunny SAT...05/17/2003...Kansas City at DC United....7:30...52 Cloudy SAT...10/18/2003...New England at MetroStars...7:30...52 Partly Cloudy SAT...05/17/2003...Chicago at New England......6:00...53 Clear SAT...04/19/2003...Los Angeles at MetroStars...7:30...54 Clear SAT...10/04/2003...Los Angeles at Chicago......7:30...54 Cloudy SAT...09/13/2003...Kansas City at Colorado.....7:00...55 Clear SAT...10/04/2003...DC United at Columbus.......7:30...55 Cloudy Columbus, strangely enough, played in both the hottest and coldest games of the year. The mean and the median were the same, oddly enough, 73.5. Most games were listed as kicking off under good conditions. The official designations were: Condition......No. Clear...........57 Cloudy..........34 Sunny...........29 Partly Cloudy...23 Rain.............5 Partly Sunny.....2 TOTAL..........150 BTW, day games (anything kicking off before 6pm) averaged 73 degrees at kickoff, while night games averaged 74. Again, nothing huge to report here, I just thought it was somewhat interesting to see the numbers.
If you can do it, one thing I'd like to see is how much weather actually affects the game. Possible correlations? I'm not sure, but here's some ideas: * temperature to scoring * temperature to shots * temperature to fouls (indicative of intensity of play?) * temperature to shot on target percent (higher = less defensive pressure?) If you include previous seasons, then there are probably enough games played in rain to try the following: * rainfall to scoring * rainfall to shot on target percent * rainfall to save percent * rainfall to fouls I think the effect of rain on the game is obvious, but it would be interesting to see how it stacks up quantitatively.
Oh man. This is actually something that could be done. That's hillarious a percent correlation with tenths of a degree to shots.
Doing those correlations would actually be interesting--it would give some support to those who defend the speed of play in MLS vs. Euro leagues by pointing to the weather (e.g., "you try running all out in a summer in Dallas or D.C.!"). I'd like to see this done!
Remember, it's only kickoff temperature, and it may or may not be terribly accurate. In any case, the closer you examine it, or the more you try to use it to correlate to something, the less reliable it may be. I, for one, would hate to try to figure out how many shots were taken when it was 79 at kickoff versus how many shots were taken 90 minutes later when it was 71, and try to ascribe it to any one factor. But I'm trying to go back and fill in temps, at least, for prior years' games to see what we have. Didn't Doug Logan have that big rant in SA several years ago where he proved (at least to his own satisfaction) that as the temperature went up, so did goal-scoring? I don't know if that's because goal-scorers like warm weather more, or defenders and goalkeepers like warm weather less.
Kenn, you didn't include me in on this little ditty? Wow, great work! Speaking as the official 'wx' girl of the boards, I couldn't have done better. Just wish I would have thought of it. Let me know if you want me to look up any past stuff. It'd be interesting to throw Tampa and Miami into the equation, once again, for the hottest - as we well know!
Well that might be the ultimate false/posative because you figure that maybe teams are gelling by late and mid season, then maybe assuming a more defensive stance when it comes to playoff time?
Don't know. All I know is this: when you assume that all other things are equal, just remember - all things are rarely equal.
Well... if we have past seasons, then rainfall might be worth analyzing, at least, because rain tends to affect play throughout, even if it stops in the first half.
Addendum: I also don't think that tracking temperature changes during a match is that important - the temperature almost always goes down during the match. My guess is that the best indicator of speed of play is not necessarily shots, but fouls. While this is not a valid comparison between leagues due to differing skill levels and differing refereeing standards, I suspect that within MLS (especially given the playing styles of many of our defenders) faster speed of play is related to defenders being more rushed to close down attackers, and thus more fouls. Actually, there is one thing that bothers me a little, which is the distinctly nonrandom effect of "who is the home team?" The preponderance of hot-weather games, for example, will feature the Dallas Burn as the home team, and most cold-weather games will feature the New England Revolution or Columbus Crew. At that point, we have to ask ourselves how much of the "weather effect" is really weather, and how much is simply a function of the teams that are playing. Nevertheless, if we can attempt to control for the teams that are playing... we still just might have enough data to work on.
Actually, I didn't think this out too well. Having just recorded all matches for this season so far (can't find last season's match reports on MLSnet any more), I notice the following probable correlations: fast play = high shot total = low shot on target percent = low foul total There are also good reasons why high speed of play should result in lower numbers of fouls being called. Slower, more tactical play may lend itself (as in Italy) to more contact and more opportunities for cynical fouling. Matches that stand out are: DAL-KC 4/24, 40 shots (highest this season), 12 shots on target (3rd-lowest % this season), 19 fouls (lowest this season) NE-LA 5/1, 15 shots (lowest this season), 10 on target (highest % this season), 44 fouls (highest this season) I'm also going to add offside calls as a possible correlation, as they tend to reflect teams pushing forward a lot.
Working on 2002, but the 2001 data is pretty spotty. Only about 2/3 of the boxscores list a gametime temperature. Team data (number of games with data/total games): Team.........Avg...Games Dallas.......87.7..(11/13) Tampa Bay....83.3..(13/14) Miami........82.2..(13/14) Kansas City..78.9..(12/13) Colorado.....77.0..(2/13) New England..76.3..(3/13) Los Angeles..72.3..(12/13) Columbus.....71.8..(10/13) MetroStars...71.8..(13/13) Chicago......70.3..(13/14) DC United....64.3..(4/13) San Jose......0.0..(0/13) TOTAL........76.8..(106/158) Monthly data (number of games with data/total games): Month......Avg...Games April......65.7..(17/24) May........70.7..(24/32) June.......81.9..(19/30) July.......83.6..(19/29) August.....82.1..(20/32) September..76.9..(7/11) TOTAL......76.8..(106/158) I think they've gotten a bit better at recording the weather in the boxscores than they used to be. Over time, the aggregate information will be interesting, and perhaps useful, perhaps not.
Kenn, pray tell, where are these box scores coming from? I (and Elninho, at least) can't find them on the MLS website. Do you have some sort of hidden cache?
In that case... can I get a listing of game-time weather conditions for the 2003 season, since you've already compiled it? Also, total shots on target and total number of offside calls for each match would be nice, if you have the time.
Weather I have. I don't have the gamestats put in anywhere yet. What am I, Jesus the Miracle Statistician all of a sudden?
I've got a bunch of crap in regards to fouls if you want it. Raw data, I came up a little short on the analysis side of it.
For 2001-2003 seasons: Monthly Avg. Temperature at kickoff Month......Avg.....G March......48.8....5 April......62.7...47 May........68.1...61 June.......79.2...61 July.......82.6...70 August.....81.9...61 September..74.8...39 October....65.2...21 TOTAL......74.6..365 Avg. Temp at kickoff by teams Team..........Avg.....G Dallas........84.6...40 Tampa Bay.....83.3...13 Miami.........82.2...13 Kansas City...78.8...40 Colorado......76.2...24 San Jose......74.1...16 New England...73.9...26 Los Angeles...73.0...41 Chicago.......72.1...42 DC United.....71.7...30 MetroStars....69.6...42 Columbus......68.2...38 TOTAL.........74.8..365 Nothing earth-shattering. I guess I didn't realize how hot it gets in Kansas City in the summer, but that's it. They've had some very high temp games. But it seems like a great many MLS games are played in decent temperatures. 174 of the 365 games in the study kicked off in temps between 60 and 79 (42 were 59 or below, 149 were 80 or above). Only 35, or less than 10%, kicked off at 90 or above. I would imagine European leagues have some hot games at the very beginning of their seasons and maybe some towards the end, but they probably have a great many cold-weather games. They also play many more games in total. It would be interesting to see the comparisons, though. 10 Hottest Games, 2001-2003 SAT...8/23/2003...Columbus at Dallas.......104..Clear SAT...7/19/2003...MetroStars at Dallas......99..Partly Sunny WED...7/18/2001...Tampa Bay at Dallas.......98..Sunny SAT...8/25/2001...Kansas City at Dallas.....98..Clear WED...7/31/2002...Kansas City at Dallas.....98..Clear FRI...8/22/2003...Colorado at Kansas City...97..Sunny WED...7/04/2001...New England at Dallas.....96..Hot SAT...6/01/2002...Los Angeles at Colorado...96..Partly Cloudy THU...7/04/2002...Columbus at New England...96..Humid SAT...8/24/2002...Los Angeles at Dallas.....96..Sunny There have been some other games just listed as "hot" with no temperature noted. I wonder if any have been as hot as 8/23/03 in Southlake, though. 10 Coldest Games, 2001-2003 WED...3/27/2002...Columbus at MetroStars....35..Cold SAT...4/06/2002...MetroStars at Columbus....38..Cold SAT...4/05/2003...Los Angeles at Columbus...41..Cloudy SAT...3/23/2002...Chicago at Columbus.......45..Sunny SAT...3/23/2002...New England at MetroStars.45..Clear SAT...5/11/2002...Los Angeles at Chicago....45..Rain SAT...4/07/2001...New England at MetroStars.46..Overcast SAT...4/27/2002...New England at Chicago....46..Showers SAT...5/18/2002...San Jose at Columbus......47..Cloudy SAT...4/27/2002...DC United at Columbus.....48..Rain What could be better than a 35 degree night in East Rutherford in March? The strange part is that Ecuador and Bulgaria played the other game of the doubleheader that night, and it drew 45,411, which will surely throw off some calculations.
8/23/03 in Southlake will probably remain the MLS record for a long time. Turf tends to get hot very quickly, so as long as the Burn's playing on grass we shouldn't see 100+ degree games with any frequency.