Club Table Counts Revisited
(Page of 7)

Three years ago I took a look at club table counts and published a summary on Bridge Winners. Two quick takeaways: (1) The top ~300 clubs (10%) represent half the table count, and (2) The typical club has a table count equivalent to a seven table game run once a week.

Since 2015, tournament table count has fallen precipitously, Jeff Bayone has bemoaned the closure of "hundreds" of clubs, Richard Willey has expressed interest in analyzing club data, and Donald Mamula has commented that the data is all there. Donald is right in principle though some old fashioned page scraping is needed to assemble all the data in a manner convenient for analysis. I'll present some results here and provide a link at end to 10 years (2010-2019) of club data for anyone who wants to take the analysis further.

Note: online clubs, e.g. BBO, are excluded from this analysis. However, the roughly 100 unit run games are included.

Let's start with total table count which has fallen 8% in eight years, concerning but not horrible. I don't show a data point for 2019 because I'm not sure of the extrapolation factor. I believe this factor should be either (12/11), (12/10), or somewhat in between. If the factor is (12/10) then 2019 has the same table count as 2018; if it is (12/11), the 2019 table count is projected to be 9% below 2018, an alarming drop.

I found over 4000 clubs for the ten year period. Almost half of them have existed for the entire 10 year period as shown in the second plot.

The final histogram shows how many clubs failed within the first year for each year. I exclude both 2010 and 2019 due to a lack of knowledge about the preceding or following year respectively. There is a spike in these one year wonders during 2014 and 2015, which is also reflected in the flattening of the total club table count during those years.

2015 vs. 2018

Though the total club table count declined 4% from 2015 to 2018, there is a core stability for the 2533 clubs that have survived from 2015 to 2018, whose aggregate table count has fallen only 1%. The figure above shows a scatter plot for the surviving clubs on a logarithmic scale. Honors in New York City is at the upper right on the right edge of the plot with 20,190 tables in 2015.

The scatter is larger than it would otherwise be because clubs that opened in 2015 or closed in 2018 have only part year table counts. These clubs are indicated in green and red respectively, based on 2014 and 2019 data.

The figure above shows the replacement curve, a rank ordering of the table count of each club lost and the club that replaced it. Clubs lost are being replaced by smaller clubs, such that only 57% of the lost table count is replaced by new clubs. Of some note, the second and third largest clubs lost were replaced by significantly smaller clubs. Given the logarithmic scale, the uppermost 11 data points are particularly important.

The 225 smallest club in 2015 were not replaced, representing a loss of ~12,000 tables. This loss is only represents one table per week per club. The clubs at the long end of the tail didn't amount to much. Their loss is not itself a crisis.

Fluctuation in individual club table counts

The histogram above shows the fluctuation in club table counts for the surviving clubs as the log of the ratio of the table counts, restricted to clubs operating for the entire year in both 2015 and 2018. The typical club has lost 4% of its table count over this three year period. In this histogram all clubs are weighted equally which means the losses at smaller clubs are more noticeable even though the total club table count for the surviving clubs is only down 1%. Over the three year period, a typical club has grown 50% or shrunk by a third (symmetric in log space), after factoring out the 4% shift.

Big clubs have done better, holding on to their table count as shown above. Their table counts also fluctuate less (18%).

Number of weekly events

Roughly half the clubs have only one or two weekly events. In this tally, separate events such as an open and a limited game held at the same time of the week, are considered separate events. The Bridge Academy of North Dallas holds 29 events per week and Honors holds 28. For the bars representing 7 or more events per week, the number above the bar indicates how many clubs have that number of weekly events.

Clubs with more events per week on average have more tables per event. Here I exclude the eight clubs with 20+ events per week due to poor statistics.

Sessions per player per year by district

In 2015 I looked the average number of club sessions per player per year broken out by district. I noted then that the cold Canadians in District 1 sure played a lot of club bridge. This is still true. Most districts have seen a drop in the average number of club sessions. Of some note the large Disttrict 9 (Florida) has dropped. District 8 (Illinois, except Chicago) has dropped a lot. District 12 (Michigan and part of Ohio) is up.

Note: This plot incorrectly relies on the 2015 district population figures for both axes. I pestered two district directors for 2018 district population but I haven't received any numbers yet. Therefore, some points may be shifted on this plot simply due to changing district populations.

I have posted 10 years of club table counts, 2010-2019, in a zip archive. Each year is provided both as an Excel file and as a tab delimited text file. The 2019 data is incomplete. It probably includes the table count through either Nov 1, 2019 or Dec 1, 2019. In early 2020 I'll update the archive to reflect the full 2019 table counts.

The columns in each file are: Rank, Club Name, Club Number, City, State, ZIP, Country, District, Unit, Tables, #Events (nEv), Club Manager, Club Homepage, ACBL Club Profile URL, and a list of weekly events. Clubs are ranked by table count in descending order. The ZIP column is a five digit zipcode for the U.S. and Mexico and a postal code for Canada. #Events is the number of events the club runs per week. Different events, e.g. an open and a limited pairs game, in the same time slot, are considered different events. The Club Manager is not necessarily the club owner, but rather the person responsible for submitting club results to the ACBL and paying the sanction fees. The club homepage is the club's website. The final column is a comma separated list of weekly events showing the day of the week and start time. If there is an upper masterpoint limit, it is shown in parentheses.

Clubs that have closed do not have club profile pages, or at least not accessible ones. These clubs have no information for the club number, zipcode, manager, club URL, ACBL URL, and weekly events fields. Moreover, when present, these fields are based on a 2019 query and filled in for previous years in the club-YYYY.txt and club-YYYY.xlsx files and thus may not be accurate for previous years, thought the club number and ACBL URL should remain constant. Many clubs do not have a website or have not shared it with the ACBL.

The files clubs-tables-2010-2019.txt and clubs-tables-2010-2019.xlsx have ten years of table count data (2010-2019) for all clubs operating at any time during that period. The nYears column is the number of years a club was in operation. Ave Tables is the average number of tables over the years of operation, rounded to the nearest integer. The file is sorted by Ave Tables in descending order.

Technical notes

Club table counts are available from the ACBL website. Here is an example for 2019:

https://web3.acbl.org/club-table-count/?year=2019&type=R

For clubs that still exist I use the club number to match data across different years. For clubs that have closed, I don't have a club number. Things get messy at this point because there are some duplicated club names in different parts of the country and some clubs have their names spelled slightly differently through the years. For the closed clubs, I match by combination of club name and unit number. Then I examined the club names and added special cases to treat ~30 clubs as matching other clubs, accounting for unit boundary shift over a 10 year period and other special cases, e.g. Ironwood Bridge, Ironwood Bridge Club, and Ironwood Country Club, are all different names of the same club in different years.