When I was young I liked to read, but I didn’t read chapter books like other kids. I read a variety of sporting almanacs and what drew me to them was the endless stats. I love to watch sports but I might enjoy reading statistics from games and trends even more. This is probably why I absolutely love the Toronto Blue Jays even though I didn’t grow up playing baseball, watching it or even reading about it. Once I learned about OPS+ and FIP it was love at first sight.
Naturally as a swimmer, stat nerd and scientists I wanted to be able to quantify my swimmers’ improvements. When I first started to coach I developed a huge rubric, probably double the size that is explained on the website. It was a great starting point for me, I would print them out once a quarter, grade my athletes and look for trends over the group to see where I as coach needed to devote more time. Additionally we could find drills and focus points for each individual athlete. I would go over those points with the athlete and discuss how to improve. Sometimes we had underwater video but often it was just a quick chat. It was around that time I learned about the computer based kinetics program Dartfish and was fascinated by its ability to measure angles.
When we started the Swim Lab, with an endless pool, I was so excited by the possibility to measure athletes quantitatively. I spent many hours fretting over the reproducibility of the results, taking multiple measurements through a session, fixing the camera positions and angles and having a consistent method of determining the start and end point for taking a measurement. It was great; a huge dedication of time and thought but the data coming out was fairly accurate and consistent. Although reliability was stable we still had areas where there could be error sources from practice to practice. Cleaning the cameras (replacing them just so), blending two video sources together, and human error of the measurement all became things we needed to be aware of. Eventually, we stopped taking the measurements, as the benefits the athletes were receiving from knowing the angles wasn’t worth it. And most athletes' feedback was the coaching was much more important to them than the angles.
Last week as any good online business would do, I was investigating some of the other swim websites to determine what type of marketing they were doing and a lot had very fancy looking videos with pause points and measurements of people’s elbow and shoulder angles. Those videos were amazing but it made me cringe and swear a few times under my breath. Because they were measuring angles from user inputted videos, the camera angles weren’t fixed, the distance from camera to athlete wasn’t fixed, and the starting point of the angle was not fixed. From someone who spent many hours watching the same clips to get as close to perfect as possible it frustrated me.
Why is this a problem? Swimming occurs in 3D but filming occurs in 2D, so everything that we watch on film is a 2D projection of a 3D motion. When you are watching athletes on film it takes a long time to understand (without consciously thinking about it) how the camera position plays with what you are watching.If you are watching an athlete swim by a fixed camera the elbow angle will be dependent on where the athlete is in relation to the camera. For example, if the athlete is swimming towards the camera the elbow angle will be greatly underestimated compared to what the actual angle is. There is a great layman’s article about this at a baseball training academy if you want to read a more detailed explanation; found here.
So should we measure angles of swimmers? Yes I think at the elite level I think there is without a doubt a benefit to determine those numbers, but the only way to do this reliably is use a 3D method or a very careful 2D measurement. Is this possible to compare these results from user inputted videos? Nope, unfortunately I don’t believe there is, trust me if there was I would be the first to implement it. For 3D measurements underwater, there is only one system that I am aware of that starts at the reasonable price of $300,000. Far out of reach for any individual athlete or team.
*** Note I did run a 3D motion lab for golf and gait analysis for a couple of years. While I don’t consider myself an expert in this field, I do when people are paying for useless services!***