What is the best method to compare more than 80 rapidly changing teams? The area’s two largest summer swim leagues, the MCSL and the Northern Virginia Swim League, employ two different, complex methods to assign teams to divisions.
MCSL: The Virtual Meet
According to MCSL automation chairman Michael Lincoln, the MCSL has utilized a computerized swim-off to determine its preseason division assignments since 1997. Lincoln and his co-chair, Bob Hincke, independently run a unique Digital Basic computer program that ranks the league’s 89 teams.
First, the program creates a virtual team for each pool. The program accomplishes this task by examining lists of times from the league’s Hy-Tek database. In every event, the program breaks down a team’s top times into sets of five. Each time in one set represents a team’s fastest time in a given dual meet. A second set lists a team’s second-fastest times from each of the five dual meets, and a third set lists a team’s third-fastest times from each dual meet. The program selects the median time from each set to create a virtual team.
Once the program has finished compiling the lineups for each team, the 89 teams are ready to face off in 3,916 dual meets. The program keeps track of each team’s number of wins and number of points scored. The teams are then ranked by number of wins. Number of points scored serves as a tie-breaker.
Lincoln concedes that the system is hardly perfect. The virtual meets cannot account for swimmers aging up and other unpredictable circumstances that accompany summer league swimming. Still, he favors the purely objective nature of the computerized swim-offs. “What we do works,” Lincoln said. “The system is not designed to be 100-percent accurate. It’s just designed to make sure that the six teams in each division are reasonably competitive and that there are not too many blowouts.”
Indeed, computerized swim-offs have been relatively successful in preventing lopsided victories. This year, MCSL dual meets were decided by an average of 67.85 points.
Although the computerized swim-offs ensure parity, the division standings of past seasons reveal that it is nearly impossible to predict the exact order in which teams will finish in a division. Since 2005, the preseason virtual meet predictions have matched the final standings in only three out of 90 cases.
A winning record does not necessarily guarantee that a team will move to a higher division, but the MCSL’s virtual meets consistently reward success. In the past five years, virtual meets have seeded 92.2% of division champions (excluding Division A) into higher divisions. In the same period, a division champion has never been demoted to a lower division.
MCSL virtual meets do not utilize times from divisional championship meets and the All-Star Meet. The use of this additional data, Lincoln explains, would disrupt the logic of the program and greatly lengthen the program’s runtime. Run on a relatively fast computer, the current program takes about 20 minutes to create the 89 virtual teams and score the 195,800 virtual events.
Even after the program has successfully sorted the teams into 15 divisions, the seeding process is not complete. The automation committee allows teams to challenge the program’s results. “After we run the first iteration of the program,” Lincoln said, “there are always a few adjustments for protests and other issues that have to be adjudicated.” Occasionally, team representatives point out a factor that could unfairly skew the results of a virtual meet. Lincoln and Hincke may adjust a virtual team’s line-up to account for an ineligible swimmer or an illegal meet entry procedure.
The day after the divisional championship meets, the MCSL Executive Board meets to conduct a final review of the division assignments before the results are released to the public.
NVSL: The Seeding Committee
While the MCSL relies on its automation chairmen to monitor division assignments, the NVSL devotes a separate committee to the process. The NVSL Seeding Committee, chaired by NVSL Director Joan Olson, compares quantitative data with the opinions of team representatives to seed the league’s 102 teams into seventeen divisions.
Like the MCSL, the NVSL ranks teams in a computerized virtual meet, but Olson declined to comment on the actual mechanics of the program. The NVSL does not release the results of the virtual meet to the public. The NVSL Seeding Committee also closely examines a team’s win-loss record from the previous year.
But these statistics are only two of many factors that influence the Seeding Committee’s decision. Every NVSL team rep must fill out a “seeding survey.” The survey attempts to provide
some insight into the intangible aspects (swimmers advancing to older age groups, swimmers moving to different teams) that a computer program cannot calculate. It also asks the rep to estimate the number of USS and year-round swimmers on each team.
“We want to look at it from each team rep’s point of view,” Olson said. “We want them to be part of the process.”
According to Olson, the NVSL Seeding Committee does not favor one statistic over another when making the final determination of a team’s division. Rather, the committee considers all of a team’s statistics in a comprehensive process, she says. A team’s dual meet win-loss record, however, seems to influence strongly an NVSL team’s placement for the following year. After the 2010 season, the NVSL Seeding Committee placed 93.3% of non-Division 1 teams with four wins or more into higher divisions. But only 13.3% of teams with three wins and no teams with two wins or fewer advanced a division.
How successfully has the NVSL Seeding Committee ensured parity? Olson points to the large number of teams that have won division championships in recent years. “Ninety-five of the 102 teams have won a division title in the past nine years,” she said.
This season, the average NVSL dual meet has been decided by 60.09 points.
The NVSL Seeding Committee will reach a final decision on next year’s division assignments some time in February. The NVSL will release the final results to the general public on February 26.
Which division assignment method do you prefer? Discuss in the comment section below.
Other division assignment procedures: