Introducing SwimSwam’s new real-time converter, powered by Swimulator!
Tl:dr: this new time converter bases its results based how that time ranks against other swimmers in a given event, course, and gender. Its designed to be equally accurate on age group swimmers as it is on olympic champions.
Some sample converted times:
Simone Manuel’s 45.56 100 Free SCY converts to to 51.76 LCM
Will Licon’s 1:47.91 200 Breast SCY converts to 2:03:86 LCM
Sarah Sjostrom’s 51.71 100 Free LCM converts to 45.52 SCY
Ryan Lochte’s 1:49.63 200 IM SCM converts to 1:53.36 LCM
But why does the swimming world really need another time converter?
The time converters made by SwimSwam, Team Unify, and others all work well for most purposes, ages, and events. After using and experimenting with them all for long enough, I grew tired of certain flaws. They were all based off of times done in different courses by faster swimmers and hence they all tended to perform well for fast, but not super fast swims. None of them were appropriate for age group swimmers and all had issues when applied to record-setting times. With these in mind, I set out to create a new time converter that would be just as accurate for elite swimmers as for age groupers and to work as well for distance freestylers as for sprint breastrokers unencumbered by these flaws.
When I would like to convert a swimming time in one course to another, there are two distinct questions that I could want to answer. The first is: “If I just swam time ‘X’ in the 100 freestyle SCY, how fast would I have gone had the course been SCM or LCM?” The second distinct but very similar question is: “What time would be similarly competitive as my time of ‘X’ in the SCY 100 freestyle in SCM or LCM?”
The answers to two questions are closely related, but could differ in a couple of ways. The first obvious reason that a swimmer’s comparative ranking in one course would be different than another – even had they swum the same event that day – is that some swimmers are better suited towards certain courses than others. David Nolan, for example, is an example of a swimmer who was incredibly dominant in SCY and while still extremely fast in LCM, was never quite as good. Another reason is that different swimmers tend to compete SCY, LCM, and SCM races, so a top 1% time in each means competing against different groups of swimmers. I chose to have my time converter answer the second question: “What time would be similarly competitive as my time of ‘X’ in the SCY 100 freestyle in SCM or LCM?”
Now why would I prefer to answer this question? As a swimmer, I certainly had more interest knowing how my improvements in my LCM times would transfer over to SCY season. However, I found a need for such a time converter when building a tool to find ways to help set team lineups and to do taper predictions. To do these, I needed a way to compare strength of times across events. For example, I needed to know whether a 22.00 50-yard freestyle was comparatively faster or slower than a 47.00 100-yard freestyle. USA swimming, while they don’t endorse the idea of a time calculator, essentially does provide one with their powerpoint calculator. It just requires a bit of extra work. What the powerpoint calculator answers is: “How does a time compare to all other swimmers in that given event/age group/gender?”
In addition, its much easier to do rank comparisons across swimming courses rather than attempting to figure out what times a swimmer would actually do swimming a different course. Without knowing that individual’s history, its would be hard to know whether they might be comparatively better in course than another. Even trickier, the conversion is not always consistent across events for a given swimmer. For example, my freestyle times were always stronger LCM than SCY, but my IM times were always better SCY (long course butterfly was not my strong suit).
So why go to the trouble of creating a new time converter when using USA Swimming ‘s powerpoints to convert times across courses? USA Swimming’s powerpoints work well on most age group and international times. However, they had some issues when used for NCAA times, which is where I did most of my analysis.The figure above shows the powerpoints for the 100th place time done in each season for all thirteen individual events for D1 and D3 men’s and women’s swimmers. As you can see, the sprint events and strokes score consistently higher points than the freestyle and distance events. Perhaps this is reasonable – maybe D3 sprinters are really faster than D3 distance swimmers. For my needs, I wanted a tool that would give similar powerpoint scores across all events and genders for the 100th place times.
To go about this, I first found all the times for each age group, event, and gender registered with USA swimming. After looking at a few distributions, I found that the times were best characterized by a skew normal distribution.
Above shows a histogram of men’s LCM 100 freestyle times at age 17 along with a skew-normal fit of the data shown in green. Pretty good fit!
Given a time done for an event in a certain course, age group, and gender, I could then convert to different course to find an equivalently fast time. I did this by using the survival function for my fit in the event being swim to convert the time to a percentile rank (the percentage of time that were slower than the given time) and then used the inverse survival function of the modeled distribution of times in the course that I wanted to convert to change that percentile into a time in the new swimming course.
For example, a time of 58.00 ranks in the 79th percentile for all 14-year-old girl’s 100 freestyle SCY times. Looking at the distribution of equivalent LCM times, we find that a time of 1:05.46 ranks the exact same., in the 79th percentile. Hence, using this method, a 58.00 SCY freestyle converts to a 1:05.46 LCM for a 14-year old female.
One of the advantages of this method is that it can be generalized to any age, gender, and event. From my analysis, there are actually fairly large differences across the genders and age groups. For example, that same 58:00 SCY 100 freestyle converts to a 1:04:88 for a 14-year-old boy, almost a second faster than the 1:05.46 given by the female conversion. The other large advantage of this technique is that as long as the model of the distribution is accurate, then the time converter should work equivalently well for extremely slow as very fast times.
The big disadvantage of this technique is that it is highly dependent on having a complete data set. It tends not to work as well for events that are competed less frequently, such as the 1500 freestyle. The converter also behaves oddly when there is some bias in terms of which swimmers are swimming certain courses. College swimmers, for example, are likely to have more fast times in yards and those that do swim over the summers tend to treat their long course seasons less seriously – I know I did.
For those of you eager to try out the new time converter, head on over to: https://swimswam.com/swimming-times-conversion-tool/?type=swimulator
This new time converter should help swimmers figure out how fast they might be in different courses more accurately than before and not give as many false hopes to those like me and my SCY freestyle times! It will certainly be more robust, especially for age group swimmers that previously only had tools that were designed for adult swimmers.