Tournament Table Counts

In a previous article, Club Table Counts, I examined ACBL club play and noted that there were significant district to district variations in club play. Notably, the snowbound D1 (Eastern Canada) had a much higher average number of club session per player than even the second ranked district, D9 (Florida). In this article I turn my attention to tournament table counts and district by district variation.

Tournament types, counts, and tables

The ACBL held 1266 tournaments in 2015, including 3 NABCs (formerly called nationals), 130 regionals, 933 sectionals, and 90 sectional tournaments at clubs (STaCs). The following table breaks this down in detail:

The total table count is similar for regionals (~171,000), sectionals excluding STaCs (~145,000), and STaCs (~124,000). Most regionals are regular open regionals and split site regionals where the latter are just an open regionals split across two widely separated playing sites within a district over the same time period. Five districts ran three of these regionals, ten ran four regionals, seven ran five regionals, two ran seven regionals, and the very large D9 (Florida) ran 9 regionals. In addition, D17 and D25 each ran one senior regional and D17 and D15 each ran one Intermediate/Novice (I/N) regional (D22 started an I/N regional in Palm Springs in 2016 and it appears it will become a permanent fixture). Finally there were three cruise ship regionals, each 500-800 tables. These are not assigned to any district.

89% of the sectional table count (excluding STaCs) comes from the ~700 open sectionals. Intermediate/Novice (I/N) is the ACBL designation for the tournaments players usually see billed as Non-LM (NLM) sectionals, e.g. NLM/750, i.e. limited to non-life masters and also excluding players with more than 750 MP even if they have not fulfilled the masterpoint pigmentation requirements to become a life master. Although there are ~200 such tournaments, they have much smaller average tables counts and so represent only 8% of the total sectional table count. The ACBL encourages I/N sectionals by significantly reducing the sanction fees and allowing the use of non-ACBL directors, the latter a large cost saving. Nonetheless, the experience of the five San Diego units is that running an I/N sectional is at best break even financially and often a loss of a few hundred dollars.

Only D20 ran Senior sectionals and only D9 ran Senior/NLM sectionals, the latter being an I/N sectional restricted by age. This paucity of senior sectionals is just as well, the term “senior” being as meaningful a qualifier in the ACBL as the term “novel” in the abstract of a scientific paper. D8 and D15 held Youth sectionals in Carlinville, IL and St. Joseph, MO respectively, a notable accomplishment. It is unclear how an NC Cruise sectional varies from a Cruise sectional; all are departing from Florida, New Orleans, or Houston. The term Downgraded appears to refer to a sectional with a very low table count perhaps such that the masterpoint awards are reduced.

The Inter Club Championships, which span multiple districts, account for the bulk of the table count under the category Other. I don’t know how these events are organized.

The numbers presented above will differ slightly from official ACBL totals due to missing table counts for 33 tournaments and the treatment of half tables. 20 of the tournaments with missing results are Grand National Team (GNT) or North American Pairs (NAP) at the district or unit level. The rest of this article does not concern these Grass Roots events. Six more of the tournaments with missing results fall in Other category and were not detailed above: UJA Federation of NY Women’s Philanthropy Bridge Tournament, Canada Wide O.F.G. #1, Mexican Bridge Federation Pair Championship Game, Rookie Newcomer Game - Shannon Oaks Senior Living, ACBL-WIDE International Fund Game 3, and the Junior USBC. The remaining 7 missing results are for STaCs which may have a non-negligible impact on the STaC results below.

Regional table count distribution

The distribution of regional table counts is shown below.

The very large Gatlinburg, TN regional is way off the right side of the histogram. At 8649 tables, it has effectively become a fourth NABC, rivaling the table count of poorly attended NABCs, e.g. Denver in the fall of 2015 (9058 tables), Providence, RI in the fall of 2014 (9574 tables), and St. Louis in the spring of 2013 (9661 tables). The Palm Springs regional (technically Rancho Mirage) is the second largest at 3876 tables, followed by regionals in Atlanta, Palmetto, FL, Houston, and Las Vegas. The average open regional has ~1400 tables but Gatlinburg pulls up the average. Excluding it drops the average to 1360 tables.

The distribution of split site regionals is superimposed (not stacked) in red. These smaller regionals average only 684 tables. Districts 2 (Ontario & Manitoba), 4 (Eastern Pennsylvania), 8 (Illinois, excluding Chicago), 11 (Kentucky and Ohio), and 13 (Chicago and Wisconsin) each ran a site regional in 2015, which show up here as 10 regionals because the ACBL tournament listing has a separate entry for each site of a split site regional. Both club and tournament play tend to be lower in these districts though D2 has high club attendance. If the two sites of a split site regional are treated as one regional, the average table count is 1368, essentially identical to the open regional table count when Gatlinburg is excluded.

Here is a list of the 2015 split site regional pairs and the Google Maps estimated driving distance and time between the respective sites:

D2, Barrie, ON & Thunder Bay, ON, 719 miles, 14 hours driving
D4, Henrietta, NY & Camp Hill, PA, 241 miles, 4 h 17 m driving
D8, Effingham, IL & Crystal Lake, IL, 252 miles, 4 h 4 m driving
D11, Owensboro, KY & Cincinnati, OH, 205 miles, 3 h 15 m driving
D13, Eau Claire, WI & Milwaukee, WI, 246 miles, 3 h 44 m driving

Sectional table count distribution

The distribution of sectional table counts is shown below.

There are handful of large sectionals. The five largest with over 800 tables each are the size of a smaller regional. These sectionals are in Atlanta (2), Roswell (near Atlanta), Houston, and Santa Clara. The long tail of large sectionals with over 400 tables is led by Texas (8 sectionals, 4 each in Houston and Richardson), Colorado (4), Arizona (3), Georgia (3), and North Carolina (3), locations predominantly in the southern half of the United States. The average open sectional has 182 tables but when the long tail starting at 400 tables is excluded, the typical sectional averages only 141 tables. Intermediate/Novice sectionals are small, averaging only 62 tables.

Frequency of tournament play

The frequency of tournament play varies considerably by district. The scatter plot above shows the average number of tournament sessions per player per year in 2015 for each district. The calculations assume that the players are members of the district where the tournament is held. For sectionals this is true with few exceptions. For regionals it is largely true but there is a group of pros and tournament regulars who compete in multiple districts.

The districts in the upper right quadrant have the most prolific tournament players. Many of these districts—those indicated in green—form a geographic arc from Texas (D19) westward and up the west coast into Canada. By contrast the districts in the U.S. northeast (red) exhibit low tournament play. The generally midwestern districts (blue) show similar collective behavior, a preference for sectionals over regionals, perhaps owing to the difficulty of travel to regionals.

D22 where I reside exhibits a preference for regionals over sectionals. I think this is due to the proximity of San Diego, Irvine, and Riverside to each other, each city home to a regional located in a major population center. Proximity in conjunction with the 10:00 am and 2:30 pm tournament schedule makes day trips for locals easy and affordable. The popular Palm Springs regional is further away but it is easy to play for N days at the cost of N-1 days of lodging, the host site lodging is reasonably priced, and cheaper options exist fairly close. Moreover, southern Californians are competitive and fairly well off and the tournaments are well run.

Outliers

The pink districts are the outliers that do not fit well into the previous groups. This collection includes D7 and D10, the two districts that form the classic American South. Curiously, they have very different participation in regionals which is something of mystery. But the evidence suggests that D7 is doing something right. Up 40% in membership since 2004 (see plot below), D7 is the ACBL’s fastest growing district by far. Gretchen Smith, a past president of D7, explains the growth as the result of hard work, a combination of teaching program, a commitment to youth bridge, and an emphasis on goodwill, and an atmosphere that makes bridge fun. After the initial publication of this article, Bob Heller, the D7 national board representative, contacted me and noted in part that D7 and D6 had long been united as the Mid-Atlantic Bridge Conference (MABC), though they are now in the process of splitting in all ways, having begun running their own regionals in January 2015.

Nearby D9 (Florida) acts like D10 in regional participation but also has fairly low sectional participation. One explanation may be that even with 9 regionals and 85 sectionals, not enough tournaments are being provided for its ~18,000 members (as of the summer 2015). If so the problem is likely getting worse because the D9 membership jumped to ~20,000 just nine months later. Another explanation may be the average age of D9 members, 2½ years older than the already very high ACBL-wide average.

D1 (Eastern Canada) is a different story. This district has 70% high club participation than average. Presumably the difficulty of traveling to tournaments, particularly regionals, in this geographically very large district limits tournament participation. D18 (Northern Rockies - U.S. and Canada) may be similar, also another geographically very large district.

A tale of four cities

Bridge is suffering in the largest metropolitan areas. D24 (New York City) tournament play is very low, though the scatter plot puts it in the worst light because D24 strongly prefers STaCs over sectionals, as demonstrated later, perhaps owing to the higher cost of tournament venues. Nor are the D24 players simply heading to tournaments in nearby districts because tournament play is also low in D3 and D25. Although D23 (Los Angeles) is presented as an outlier it may have something in common with D24, likely expensive venues compounded by miserable commutes. Some Bridge Winners members have suggested that D23’s adherence to the traditional 1 pm and 7 pm tournament schedule may be hurting tournament participation. D13, which includes Chicago, exhibits similar low regional participation.

The Washington D.C. area, part of D6, is the standout exception. Tom Herzog, one of my regular partners, recently retired from the D.C. area after many years of working for the federal government, cites the extraordinary number of Ph.D.s in the D.C. area. He goes on to suggest that the reasonable job hours of government workers may allow more time for bridge than for those engaged in say the New York City financial industry. The D6 alliance with D7 to form the MABC may also have been to the benefit of D6.

Tournament play versus club play

On a district by district basis there is no significant correlation between club play and total tournament play (regionals + sectionals).

Sectional tournaments at clubs (STaCs)

STaCs straddle the boundary between club play and tournament play. Like a charity game or special club game, the awards are based on the tournament masterpoint calculations rather than club masterpoint calculations. Like a regular sectional, the pigmentation is silver. But in a STaC, the competition is typically the usual group of club players rather than the stronger field that an ordinary sectional would draw.

STaCs can have overall awards based on the participation of multiple clubs using the same stratification in the same type of event. But unlike a tournament where an effort is made to balance the field across sections, the field may vary considerably among the participating clubs. A strong pair paying a visit to a club with a technically open event but weak field strength may well rank quite high in the overalls, particularly if the number of tables is small, leading to higher natural fluctuations in the scores of the participants.

I think STaCs got started as a way to give players who were underserved by sectionals an easier way to earn silver points. Invariably, some STaCs became fund raisers because the sponsoring organization—a unit, district, or collection thereof—can tack on a sanction fee above the ACBL sanction fee. For example, the three annual Great Western STaCs held primarily by D17, D21, and D22 are the largest funding mechanism for the monthly Contract Bridge Forum newspaper published by the Western Conference.

STaCs event used to pay the same award as regular sectionals. But the January 2015 revision of the Masterpoint Award Rules & Regulations set the rating factor (R) to 11 and 9 for ordinary sectionals and STaCs respectively. This means that if things are otherwise equal, a STaC event awards (9/11) = 82% of what a regular sectional would award.

D3 (New York and New Jersey) is a clear standout, with more than twice the STaC participation as any other district. I have no idea why. Neighboring D24 has high STaC participation and almost no sectional which might be attributed to the high cost of venues in New York City. D14 (Iowa and the Dakotas) had only one STaC, shared with D15, but the rest of the Moderate Midwest clusters in its typical manner.

Table of tournament play by district

The following table shows the numbers used to generate the three scatter plots. The Reg, Sec, and STaC columns are the average number of regional, sectional, and STaC sessions respectively per player in 2015 for each district. R+S = Reg + Sec; the districts are sorted by decreasing tournament play. nR and nS are the number of regionals and sectionals respectively held in each district in 2015. nRs and nSs are nR and nS respectively scaled to a district with the average district membership of ~6,600 players. Mem is the district membership as of June 2015.

In this table the five split site regionals are each counted as one regional because they are coincident in time. This reduces the 130 regionals in the ACBL tournament list to 125 for an average of 5 regionals per district. After the scaling for district membership size, it appears that D8 and D5 are overserved by regionals and that D21 and D9 are underserved. D20 is overserved but their level of regional participation shows that the excess is not going to waste.

933 total sectionals (excluding STaCs) is an average of 37 per district. D3 and D23 are underserved by sectionals. D24 (NYC) is starved.

Participation by gender

The male to female ratio for the ACBL membership as a whole is 37% to 63%. This is primarily because women have a longer life expectancy and that the average age of ACBL members is so high that the life expectancy difference plays a big role.

STaC participation is 39% male, only slightly higher than the percentage of male members. This probably because most players don't seek out STaC games but rather simply end up in one when their local club holds one on a day of the week that the member would ordinarily play.

Sectional, regional, and NABC participation is 43%, 44%, and 52% male respectively. Men it seems enjoy strong competition. Not a big surprise.

Gene Saxe on Bridge Winners wondered if the traditional vs. commuter regional schedule affected gender participation rates. It makes little difference, just 0.8%, with men favoring the traditional schedule (45.4% of participants) over the commuter schedule (44.6% of participants).

Analysis methods and notes

The ACBL website has a tournament listings interface. The table is populated using JSON retrieved via an AJAX query. For example, the following query will return JSON for all 2015 tournaments, which can then easily be parsed programmatically:

http://tournaments.acbl.org/includes/ajax/tournamentslist.php?year=2015

Unfortunately neither the data table nor the JSON retrieval includes table counts. But they can be obtained by programmatically following the Results link for each tournament. For STaCs the Results link mistakenly leads to the tournament schedule rather than the tournament results. However it is possible to follow the Masterpoint Winner List link from the tournament schedule to arrive at the STaC tournament results page. On the tournament results page, the table count is available from the line “Total attendance: # tables” line generated by ACBLscore.

The number of players attending is also not provided. But the number of players winning masterpoints can be parsed from the line, “Total masterpoints: # earned by # players”. The number of masterpoint winners is fairly close to the number players because attending because about 50% of the participants in any event earn some masterpoints. And Swiss team events award small match awards; a team would have to lose every round to end up with no masterpoints. Based on the examination of tournaments for which I have all the ACBLscore game files, 90-95% of regional participants, ~80% of sectional participants, and ~70% of I/N sectional participants earn some masterpoints. These figures can be used to estimate the number of participants at each tournament given the number of masterpoint winners.

The ACBL is missing results for some tournaments. For 2015, this primarily affected GNT/NAP Unit and District finals and a number of one-off tournaments lumped under the Other category. Neither problem affects the analysis. Unfortunately seven of the missing results were STaCs. There was one missing result from each of D1, D6, D9, D13, D17, and D23.

Some NABC results are missing too. But NABC table counts are tracked elsewhere, even on Wikipedia, and I have simply hard coded the NABC table counts in my Perl code.

The total attendance shown in the ACBLscore reports never seems to include half tables. This issue and the missing tournament results will lead to minor discrepancies between the results presented here and official ACBL reported tournament total table counts.

Occasionally tournament starts in one year and end in the next. For this analysis I examined all tournaments starting in 2015 regardless of end year.

Some STaCs involve multiple districts even though only one district appears in the tournament table (and JSON retrieval). In these cases, I have apportioned the table count as well as possible. Specifically, I apportioned it according the number of participating clubs from each district which again can be programmatically queried. Better apportioning might be achieved by mapping the city associated with each STaC masterpoint winner to the appropriate district but this approach is complicated, still not perfect, and unlikely make a significant difference in the analysis.

Here is a sample URL for the list of participating clubs in a Great Western STAC where 1508005 is the sanction number

Note: This query takes about 30-60 seconds to generate the table with the list of participating clubs and no busy indicator is provided as on other ACBL web pages that fill a table with a background AJAX query.

Gender ratios are based on the number of tournament participants. Weighting by the number of sessions seems more appropriate but is not possible with the data available. Moreover participants here is actually the masterpoint winners rather than all participants. But even if there is a gender skill difference on average, the multiplicity of events and strats within events should largely equalize the chance of scratching as discussed above.

Get the data

Several datasets are available for download as zip files. Each zip file includes the data in both tab delimited text format and in Excel format, both well suited for further analysis.

Tournament Breakout for 2015
Tournament Statistics by District for 2015
District Membership 2004–2015
Tournament data for 2015
Tournament data for 2003–2015

The first file is the table presented in the section Tournament types, counts, and tables.

The second file is the table presented in the section Table of tournament play by district.

District membership numbers are taken from the month 6 (June) QUIP report for each year. Note: the data for 2005 is missing.

The fourth file contains the primary data used to produce all of the preceding tables and plots in conjunction with the June 2015 district membership numbers which are available in the previous file. The column nWin is the number of masterpoint winners, a surrogate for the number of tournament attendees. The Tourney Id column is the unique ACBL tournament identifier, either actually or effectively a database primary key. The cells are hyperlinked to the corresponding tournament schedule page on the ACBL website. The HTML column gives the base name of the tournament results page and the cells are hyperlinked to the corresponding tournament results page on the ACBL website. The %Male column is percentage of participants—strictly speaking masterpoint winners—who are male. The Start Time Frequencies column gives all start times used in the tournament and the number of events with each start time. For examples, “10:00 am (9), 2:30 pm (11), 7:30 pm (8)” means 9 events started at 10:00 am, 11 started at 2:30 pm. For ease of analysis (in particular sorting), the Top Two Times column shows the two most frequently occurring start times, i.e. "10:00 am, 2:30 pm" in this example, time sorted. This column can be used to distinguish between the 10:00 am / 2:30 pm / (7 pm ghost town) or nearly equivalent “commuter” schedule that has become standard in D22 and the (9:30 am diehards only) / 1 pm / 7 pm traditional schedule.

The final download is a 13 year dump of all ACBL tournaments. Hopefully it will stimulate additional analysis. Be aware that tournament table counts are probably undercounted during the period March–September 2006 and are certainly undercounted prior to March 2006 (except NABCs). This is because the ACBLscore report used to generate the tournament winners and list of masterpoint winners didn’t include the “Total attendance: # tables” line prior to September 2006. From March to September 2006 tournament table counts are computed by adding up the tables from each event, i.e. the numbers in lines such as “Stratified 199er Pairs - 8.0 Tables” Note: care must be taken when parsing lines such as “Stratified Open Pairs - 29.0 Tables / Based on 37 Tables” This means the event had 29 tables but the masterpoint where based on 37, i.e. the open event was allowed to include the tables from the simultaneous limited event. Prior to March 2006, the size of the events was reported as either the number of pairs or teams rather than the number of tables, e.g. “Friday Afternoon Pairs - 74 Pairs”. It is easy enough to divide the number of pairs by two when tallying. But when totals are presented in terms of pairs or teams we need to adjust for multi-session events. For example, a two session Swiss with 40 teams should be counted as 80 tables. Similarly, Knockout Team events have subsequent rounds of play with diminishing numbers of participants. It would be possible to write more sophisticated code that made reasonable assumptions based on event names and also the number of masterpoints awarded but this is tedious and not done in the current code, hence the undercounting.