In this post, we’ll examine statistics related to USA-S competition format. We’ll see that older, faster athletes are more likely to compete in the “preliminaries and finals” format than younger athletes, resulting in an age-related bias towards fewer and faster swims per older athlete.
USA-S recognizes two competition formats: “timed finals” and “preliminaries and finals”. In a “timed finals” event, athletes compete once and their times determine finishing places for the event. In a “preliminaries and finals” event, participants first swim the event to determine who may compete in the finals. Then the fastest swimmers from this preliminary round swim again in the finals, with their times in the finals determining the top finishing places for the event.
Three quarters of age group swims are from timed finals, where each participant swims an event exactly once.
While February and November have the most age group swims, March (end of short course season) and July (end of long course season) have the most preliminaries and finals swims. The beginning of the short course (September/October) and long course (April/May) seasons have the fewest preliminaries and finals swims.
We’ve already seen that age group participation peaks at age 12 and declines thereafter.
After age 12, the frequency of timed finals swims declines precipitously while the frequency of preliminaries and finals swims remains nearly constant.
As shown by the next plot, the age-related decline in timed finals swims dramatically decreases the fraction of timed finals swims for older athletes, from 98% at age 6 to 43% at age 18. More than half of the 17/Over swims are in prelims/finals format.
The fraction of athletes who never participate in a timed final swim also increases with age. 25% of age group athletes will only swim preliminaries and finals events in their 18th year.
This age-related decline in timed finals swims means that older athletes are able to swim fewer events per day on average than younger athletes. USA-S rules allow age group athletes to enter at most six timed finals events per day or three prelims/finals events per day. An athlete entered in three prelims/finals events may swim at most six times (if they make finals in all three events) or as few as three times (if they don’t make finals in any event). We’ll confirm this age-related bias in a another post.
The “preliminaries and finals” format allows the fastest athletes to swim their best events twice per meet, which biases our data towards faster swims. The bias increases with athlete age because older athletes participate in more preliminaries and finals events.
The following plot shows the mean rank of times achieved in the two competition formats. Lower mean rank implies lesser (ie., faster) times. Conversely, greater mean rank implies greater (ie., slower) times. The rank of a swim is determined relative to other swims in the same event by athletes of the same gender and age in years. This plot confirms that the preliminaries of prelims/finals events contain faster swims than timed finals events on average, which strongly suggests that the “preliminaries and finals” format is more likely to be used for events with fast qualifying times, such as championship events.
Please note this plot does not mean that a given athlete will swim faster in preliminaries than timed finals. It simply means that the swims in a preliminary are faster on average than the swims in a timed final. Next we’ll learn this is because preliminaries typically have qualifying times, which explicitly exclude slower athletes.
Roughly 50% of the “preliminaries and finals” events from our LSCs have qualifying times, while less than 10% of the “timed finals” events have qualifying times. The fraction of events with qualifying times does not vary significantly by age or gender.
“Preliminaries and finals” events have three stages: preliminaries, swim-offs, and finals. Swim-offs occur when multiple athletes tie for the last slot in the finals, or for one of the alternate slots in the finals. The winner of the swim-off earns the contested slot. Swim-offs are the rarest occurrence in age group swimming, comprising less than 0.006% of all swims.
The ratio of preliminaries swims to finals swims is 2.5, meaning there is on average one finals slot for every 2.5 preliminaries slots. In other words, events with one finals heat have on average three preliminaries heats, while events with two finals heats have on average five preliminaries heats. Swim-offs account for only 0.03% of preliminaries with finals swims.
The depth of preliminaries varies from month to month. Preliminaries held in October are under-subscribed at one finals slot per 1.9 preliminaries slots (2 preliminaries heats), while those held in December are more heavily subscribed at 2.9 (3 preliminaries heats). Many LSCs sanction their first championship meet of the short course season in December, when short course training first peaks.
Swim-offs are most likely to occur in June and November, when age group competition peaks but preliminaries with finals are slightly undersubscribed.
The following plot confirms that finals swims are faster on average than preliminaries swims. Again, this plot does not imply that a given athlete will swim faster in a final than in a preliminary; it merely confirms that finals are restricted to faster athletes. In our next post, we’ll see that athletes who make finals are in fact more likely to swim faster in finals than they did in preliminaries (link).
The preliminaries and finals format collects the fastest athletes from a large pool of entrants into a single heat where they can all race against each other for a fixed number of slots. It primarily benefits spectators at elite meets, by allowing them to see the fastest athletes race each other head-to-head in a single heat. It also arguably results in a more accurate outcome, by allowing the fastest athletes of the day to race head-to-head against their competitors in the finals for a fixed number of slots, irrespective of seed time accuracy.
As such, the preliminaries and finals format is appropriate for situations where there are a large number of competitors and a tiny number of meaningful finishing rewards. For example, the 2016 USA Olympic trials had 180 women competing for only 4 Olympic team slots in the 50 Free (22.5 prelim slots per final slot). At the 2016 Olympics, 90 women competed for only 3 Olympic medals in the 50 Free (11.25 prelim slots per final slot).
Age group swimming is nothing like that. Firstly, the number of competitors in age group events is relatively low. On average, preliminaries and finals events have only 2.5 prelim slots per finals slot. Some events have fewer entrants than finals slots! Secondly, the finishing slots of age group events are not particularly meaningful because the fastest athletes in the event are mostly those at the top of their age group as of the event’s age-up date. If an event’s age-up date were adjusted by even a month, the top finishers would be different. Thirdly, the reward for a top finish in an A final is modest - a medal or a ribbon. The top finishers in the B and C finals receive no awards; swimming those finals merely ensures that they won’t lose positions in the overall event ranking. In age group swimming, athletes compete primarily with themselves, to achieve better times that qualify them for more more prestigious meets. And finally, spectators have little interest in returning to the pool at night to observe a finals heat.
Preliminaries and finals events come at a significant cost to both age group athletes and their families. Athletes who don’t make finals are denied the opportunity to compete in up to three additional events per day. Athletes who make finals are expected to return to the pool for a one hour warm-up session followed by two hours of competition. Families are required to provide officials, lane timers, safety marshalls, gate workers, hospitality, and so forth. Preliminaries and finals events turn a six-plus hour meet commitment into an onerous dawn-to-dusk commitment for the athlete and their families. Athletes who scratch finals may find their teams disinvited from future age group meets.
The inappropriateness of the preliminaries and finals format for age group swimming is compounded by circle seeding, which distributes the fastest seed times among the last three preliminaries heats. When the last three heats have similar seed times, circle seeding does not materially delay the meet. But when an event only has three heats, as frequently happens in age group meets, spreading the three slowest seeds across those three heats can delay the meet significantly because every heat is delayed by its own slowest athlete, which can be cruelly embarrassing to the slowest seeds.
To address this unreasonable situation, we recommend that USA Swimming limit the preliminaries and finals format to open events with at least 8 registrants per finals slot; and that alternate finals slots be prohibited for 18/Under athletes. Events planned as preliminaries and finals that do not meet the participation quota would be automatically converted into timed finals at the meet.
In this post we learned that competition format creates three distinct data selection processes. Timed finals contain a representative sample of swims by younger athletes throughout the season, but are missing some older faster athletes. Preliminaries contain a sample of faster athletes at the peak of their training (towards the end of the season). Finals contain an additional sample of the fastest athletes at the peak of their training and the top of their age group.
We’ve also seen that older and faster athletes are increasingly likely to participate in “preliminaries and finals” events rather than “timed finals”. This has three effects: it decreases the mean number of swims per athlete, it encourages athletes to specialize in their best events, and it biases our data for older athletes towards the fastest swimmers and fastest swims.