WOD Data: Crossfit Open 15.4

15.4! Hand stand pushups and (pretty) heavy cleans.  Did you have em?  More people showed up this week with HSPUs than had muscle ups last week, but it was still a workout that pushed a lot of athlete’s limits.  Somewhat surprisingly, we’re seeing almost no attrition since the first week.  I think you could argue at this point that having the scaled division is keeping people in the game while allowing for more challenging Rx programming.

I’ve added 15.4 to my public collection of Open data, so feel free to download and see what you can add.

crossfit_15.4_participation

15.4 Rx

8 minute AMRAP of
3, 6, 9, 12, 15, 18 handstand push-ups
alternating with
3, 3, 3, 6, 6, 6, 9 cleans (185/125)

5,975 athletes chose Rx but didn’t have the cleans and took their 3 reps to the top Rx club on the leaderboard.

Congrats to the 4,643 athletes (~3k men, 1.5k women) who got scores of at least 4 even though the clean & jerk PR on their profile was less than 185/125.  You either rose to the Open occasion and rocked it or you’re bad at updating your profile.

Many more athletes ran out of time during HSPUs than cleans (controlling for the difference in total reps).  That means people were knocking out the cleans faster which is surprising to me as one of the people in that big 3-rep, no cleans spike.

crossfit_15.4rx_hist_gender

No huge jumps in the percentile plot this WOD, aside from the first clean for Women.  In general each rep mattered just about as much as the last.

crossfit_15.4rx_percentile_gender

15.4 Scaled

8 minutes AMRAP of:
10 push presses (95/65)
10 cleans (115/75)

Because there was no big difference in the difficulty of the movements, we get a nice normalish distribution in the Scaled division.  The exceptions are the big spikes at the end of each movement.  That’s presumably part psychological, but also reflects the time it takes to load and unload the bar.

crossfit_15.4sc_hist_gender

Again, remarkably smooth percentile plot which means it was more about your “engine” than strategizing around specific numbers. Slight exception were 81 and 101 reps which were worth a few percentile points in a single rep.  I bet the first 50 reps felt like more of an accomplishment than just getting out of the bottom 10%, but that’s what happens when everyone can do them.

crossfit_15.4sc_percentile_gender

Age

Have you ever heard that strength peaks in the mid-20s, but endurance athletes peak in their 30s? Have you ever seen a graph that illustrated it so clearly?

crossfit_15.4_age

Height & Weight

If you have to invert yourself and push (Rx), you don’t want to be too tall or heavy.  If it’s a question of high reps on a fixed, relatively light weight (Scaled), you want to be bigger than the weight.  By as much as possible.

crossfit_15.4_height

New plot! In previous weeks, people have asked about whether it’s really about height or weight since they are obviously correlated. The answer is interesting.  For Rx, being on the short side was an advantage, but there was a sweet spot of weight.

For Scaled, heavier athletes did better at all heights, BMI be damned!  115/75 cleans are doable for most but for those who weigh more than twice that, it probably came down to how fast you could change your weights.  Men over 240lb averaged 100 reps and 70th percentile ranks. crossfit_15.4_height_weight


– Sam Swift.  I’m a data-scientist at Betterment, and 3 years into CrossFit, now at CrossFit Prospect Heights (previously CrossFit Carrboro).  I’ve got more posts in the works, follow me @swiftsam. All 2012-2015 data available for download here. Scraping and visualization code available here on github.

Posted in Posts Tagged with: ,
  • http://www.residentassistant.com danolie

    Great analysis!!! How did you get the raw data to work from? (I have wanted to figure out something as simple as my percentile ranking at age and the data is hard for me to find beyond the basic leaderboard)