There was some lively conversation in the preseason thread about what the Colorado numbers meant. Well I read my horoscope, consulted my tea leaves, and interpreted the smoke rising from the TA in the Monday night tilt. And while none of them could tell me the meaning they all assured me that the conversation would not be rehashed here in the week 1 thread. All other numbers however are up for discussion. Remember in this thread we discuss the: (AAQ = Attendance Analysis Quotient. This figure is an overall attendance assessment, calculated from the weighted values of average attendance, median attendance, <10K percentage and >20K percentage. A lower figure represents a better attendance performance.) This is a slight change from Andy's old formula which used simple ordinal values and provided a full ranking point separation for two numbers that may have been within a percentage. This ordinal system did not allow for fine differences to be seen, and may have painted a somewhat skewed view of the numbers (though to be honest there is not much difference in the two end of season sets). The formula to find the weighted value for each column is: Take the annual value and subtract it from the Max value for it's column. Then take that result and divide it by the (Max-Min) for its column. For example, for the average column 17410 is the max 13756 is the min 3654 is the difference To calculate the derived value for 2010 Subtract 16675 from 17410 to get 735. Divide 735 by 3654 then mulitply by 100* to get 20 *Note that this step is new as some people thought whole numbers would be easier to read. MLS Attendance is based on tickets distributed not tickets scanned. Therefore if a team sells or distributes X number of tickets, then X is the attendance for that match. Even if 0 people showed up for whatever reason, X would still be the official attendance. There are a number of reasons actual attendance may be lower or higher than the announced. As I found myself telling NLSANAND, I no longer have a huge issue with people discussing their perception of the attendance at a match. However, and this is a big however, if the discussion again breaks down to the reported numbers are just plain wrong the hammer will come back. Feel free to discuss the perception and why it may not seem like what was reported, do not however insist that the numbers we use are flawed, because they are after all the only numbers we have. I think we all understand at this point why the reported number may just not be what was expected. Let's give this a try and see how it goes.
So I am just kinda getting back into the swing of this. Not quite in mid season form yet. Here are the numbers for this week. Opening week saw eight matches starting with kickoffs in Colorado and Vancouver, and culminating on a wet night in Portland. Last Weeks Games: Code: Colorado 14746 Vancouver 21000 DC United 16314 San Jose 10525 Los Angeles 27000 FC Dallas 20906 Chivas USA 14464 Portland 20438 [B]Total 145393[/B] [B][COLOR="Red"]Average 18174[/COLOR][/B] Week 2 will see a total of nine matches. With the inaugural match in Montreal where over 50k are expected. Also kicking off their home season will be Seattle, Philadelphia, Real Salt Lake, and Sporting. While San Jose, FC Dallas, Chivas USA, and Los Angeles all get their 2nd home games. I am not going to start the predictive numbers until 1/3 of the way through the season, so until then milestones, and comparitive will go here. There were not any milestones reached in week 1. The comparitive numbers for what they worth are as follows: Comparison to This Point Last Season and All-Time Averages: Code: ----Team---- Current Last Diff Alltime Diff Chicago 0 0 0.0% 15529 0.0% Chivas USA 14464 18122 -20.2% 15863 -8.8% Colorado 14746 17139 -14.0% 13971 5.5% Columbus 0 0 0.0% 15238 0.0% DC United 16314 18132 -10.0% 17194 -5.1% FC Dallas 20906 20145 3.8% 12203 71.3% Houston 0 0 0.0% 17327 0.0% Los Angeles 27000 27000 0.0% 22067 22.4% Montréal 0 New 0.0% New 0.0% New England 0 0 0.0% 15610 0.0% Philadelphia 0 0 0.0% 18724 0.0% Portland 20438 18627 9.7% 18827 8.6% Red Bull NY 0 0 0.0% 17159 0.0% Real Salt Lake 0 0 0.0% 16828 0.0% San Jose 10525 10525 0.0% 12820 -17.9% Seattle 0 0 0.0% 35329 0.0% Sporting KC 0 0 0.0% 11386 0.0% Toronto FC 0 0 0.0% 20264 0.0% Vancouver 21000 22592 -7.0% 20412 2.9% Overall 18174 21616 -15.9% 15739 15.5%
Last but not least will be the AAQ numbers. Not a bad initial week, and the big hosting of the Impact and Seattle could drive week 2 umbers higher. Per request I have added the column header as a footer row as well. Code: Current Average Median <10k >20k AvgPts MedPts <10kPts >20kPts AAAQ AAAQRnk Date 1996 26141 25870 0.0% 100.0% 0 0 0 0 0 1 4/13 1997 15000 13151 0.0% 25.0% 91 98 0 75 264 12 3/29 1998 19723 16122 0.0% 40.0% 52 75 0 60 188 6 3/21 1999 16043 16872 0.0% 0.0% 82 69 0 100 252 10 3/20 2000 14619 12891 20.0% 20.0% 94 100 80 80 354 17 3/25 2001 20192 19801 0.0% 50.0% 49 47 0 50 145 4 4/7 2002 14484 13234 0.0% 0.0% 95 97 0 100 293 15 3/23 2003 19336 22231 25.0% 75.0% 56 28 100 25 209 8 4/12 2004 19368 22480 25.0% 75.0% 55 26 100 25 206 7 4/3 2005 13899 13484 0.0% 0.0% 100 95 0 100 295 16 4/2 2006 21657 20739 0.0% 60.0% 37 40 0 40 116 2 4/2 2007 16050 15353 0.0% 0.0% 82 81 0 100 263 11 4/7 2008 14866 13843 0.0% 20.0% 92 93 0 80 265 13 3/29 2009 16942 15979 0.0% 16.7% 75 76 0 83 235 9 3/21 2010 18035 16095 16.7% 33.3% 66 75 67 67 275 14 3/27 2011 21616 19765 0.0% 50.0% 37 47 0 50 134 3 3/19 2012 18174 18376 0.0% 50.0% 65 58 0 50 173 5 3/12 Current Average Median <10k >20k AvgPts MedPts <10kPts >20kPts AAAQ AAAQRnk Date
Week 1 has good numbers. Nothing to get too pessimistic about and nothing to get too thrilled about. Just a solid week. Week 2, however, should be a killer week. Montreal and Seattle pitching in about 90,000 between them. LA, Philly and Sporting will likely be very strong, and RSL probably as well. And isn't the San Jose a double-header game with a Mexico u-23 match? I won't be surprised if the week easily topped 20,000 average. MLS attendance is off to a good start. Thanks, as always, for posting the stats.
Yes, the doubleheader is being held at AT&T Park in San Francisco this weekend. Capacity is about 41,900. They've been promoting the event heavily in the Bay Area with billboards on the major highways. We should clear at least 25,000.
How times have changed. 18k for the week and its just solid..... In years past we would have been ecstatic.
Anyone know what the Capacity of FCD Stadium is now? I thought it was a shade over 21K before and they added additional seating on the Stage yet announced a sellout of under 21K?
While it is always nice to view the sky as the limit, keep in mind that while the season did start earlier in the year than ever before (and there were justified concerns about how the unpredictable/harsh weather in early March could/would affect attendance in some markets), the early weeks in the season will likely see some "bump" in attendance (or certainly some data points above the expected final attendance averages in a few markets) as teams play their home openers. in some markets, the home opener is the high water attendance mark for a good portion of the season (perhaps until some marketing/special event comes along or until the last 1/3 of the season when ticket sales are again usually strong and the playoff push is really on). and there are also some interesting points/counterpoints/insights on this topic mentioned in the Preseason thread, especially between and including post #161 and post #169.
something i will be keeping track of so i might as well share: Total: 145,393 Games: 8 Average: 18,174 Median: 18,376 Median-33%: 12,257 Median+33%: 24,514 <MED-33%: 1 / 12.5% >MED+33%: 1 / 12.5%
What is this median +/- 33%? The standard deviation was 5,154. The standard deviation is really easy to calculate. Though I would note this is a really small dataset to start calculating these kinds of statistics. But, for discussion, I suggest that in general, large standard deviations are bad, low standard deviations are good. A caveat would be poor attendance combined with a low standard deviation, but there really aren't many examples of that. Teams with low attendance tend to use "gimmicks" to increase attendances for "events". FWIW, the league as a whole had a standard deviation of over 11,000 in 1996. LA by itself had a mind-boggling standard deviation of almost 25,000 that year!! How the hell do you create accurate budgets for the following year with data like that?
And for the non-statisticians, measuring the average attendance plus one standard deviation and minus one standard deviation captures a little more than 2/3 of all samples (where samples are the specific attendnance totals of each game).
the Median +/-33% are what are technically speaking more accurate measures than 10K/20K for "bad" and "good" games. not to start this debate again ... the AAQ is how it is and no need to change it going forward as it would be a hassle to retroactively adjust the data. but in a league where the Median attendance is 18K measuring what games fell below 10K and what games fell above 20K isn't very useful since they do not represent proportionate things (one represents 8K less than the median the other represents 2K above the median) ... this is just a more accurate way to look at that in the context of the league Median not being 15K anymore. and you are of course correct ... it is week one ... the dataset is 8 games large ... so none of these really mean much in that context but my spreadsheet is set up to calculate it automatically as i add games each week so i just thought i'd share it for those who are interested in seeing how the "10K/20K ie "bad/good" metrics look adjusted for a league with a higher median attendance.
in 1996-1997 setting "accurate budgets" wasn't really a top priority, it would seem for MLS. the guys who started up the league knew they were most likely going to be blowing money for a few/several years (while they played in over-sized rental venues, that allowed for those huge standard deviations in attendance to exist). And, after 5 or so years, they could come out with some big number (hey the league has bled some $350M as a start up these first years), and make the appropriate decisions to contract a few franchises and move the league forward in a new and better direction (focusing on getting venues built). now, as more teams are in more appropriately sized venues, the overall business is a lot easier to understand and predict (and run and profit from).
LA Galaxy 1996 attendance by game: 69,255, 40,347, 20,471, 25,231, 15,231, 92,216, 21,280, 62,703, 17,137, 15,089, 10,251, 13,073, 19,279, 8,561, 14,179, 18,347
I can't believe I am getting sucked into this. This will be my last post on the subject. I totally understand that you want to illustrate a better way to show attendance performance. I get it. my point is the +/- 33% is stupid. You want to show variation - fine. Show attendance + or - one standard deviation.
i am not trying to show variation. that is what standard deviation is for. orignially, as part of the AAQ the person who designed it wanted to measure what proportion of games fell below and above certain measures. in other words they wanted to measure the number of games that were "bad" (some significant amount below the average/median and the number of games that were "good" (some significant amount above the average/median. back then the league median was about 15K ... so it made sense to say ... okay any game 5K over the average/median of 15K is "good" and any game 5K under that 15K median is "bad". thus was born <10K and 20K>. and that is all fine and dandy as long as your median stays around 15K. your "good" and "bad" measures are equidistance from your median so counting the number of games that fall into those categories and reporting the proportion/% is valid. BUT ... if the median goes up 3K to 18K then you are no longer using a fair or proportionate definition for your "good" and "bad' ... thus your count of them and your reporting of a % of them is unfair .... what good is it to say x number of games fell 8K below the median and y number of games fell 2K above the median? it is meaningless. so to adjust for a new median you simply redefine the "good" and "bad" thresholds as some set % of the median ... +/-33% (which is what 10K and 20K represent when the median is 15K). now no matter what the median is your buckets for "good" and "bad" are always proportionately defined so when you count them and report the % of them it is actually a meaningful metric. now, this was debated many times and i am not trying to start the debate again and get the AAQ changed. i was told that would be a lot of work and the person who is gracious enough to dedicate their time to do the data for this thread did not think it would be worthwhile to go back and retro fit all of the previous 16 seasons ... and i respect and understand that, it makes perfect sense. but by the same token there is nothing wrong with keeping track of the more accurate and valid measure and reporting it here for those who wish to know it. that is all i am doing ... it is interesting data and hopefully some people find it useful ... others may not but it is only going to be a small single post and it is in no way supposed to be some sort of new salvo in the debate of the AAQ ... that debate is closed ... the AAQ is what it is and i am fine with that.
Oh god, here we go again. You know Popgun, for somebody that complains when anyone drives your threads towards off-topic, you tend to drive a lot of threads that way yourself.
how did i put this thread off topic? i put a small post with some numbers i keep track of and thought it might be good to share. that was it ... it didn't contain any commentary it didn't contain any argument to change the AAQ it didn't contain any criticism of any of the other metrics. it was just some data i keep track of that i shared. it was other posters that attacked that and started the derail. i only sought to explain what the numbers were, how i arrived at them and what they were useful for ... if that is derailing then i guess i should just sit back and let people take a shot at the contribution without responding?
I know you were joking here but for those that don't know the history, 1999 was the first year we started to track attendances informally league wide on what was then called the SoccerBoards (before Jesse rebranded as BigSoccer). Back in 1996, fans were mostly separated into individual mailing lists based on each team.