Where did you really stack up in the 2014 CrossFit Open WODs?

If you’re like me, you know your scores on the  CrossFit Open WODs, and you looked up what ranks those scores got you.  But what does 71,786th out of 153,305 mean?   Not much. The games website lets you drill down some, but I need to see it all.  I collected all 802,164 scores from over 150,000 athletes in this year’s games and took a look.

14.1

To refresh you memory …
10 minute AMRAP
30 double unders
15 power snatches (55/75)

crossfit_14_1_annotated

(click on any of the graphs to zoom into a full res version)

The first thing you see are huge spikes at the end of each set.  The weight didn’t go up in 14.1, so these aren’t lifts people got stuck on.  There are probably two main reasons people end up at these very popular scores.  Some people run out of time as they’re trying to switch from the rope to the bar or back. The bigger reason is probably that people push to finish round numbers and complete sets.  It feels good to finish that last snatch and collapse on the floor with a 225, but squeezing in even one double under could set an athlete apart from that 2,943-way tie.  Also, it looks like 424 people spent their 10 minutes and succeeded in getting (probably) their first double under.  Congrats!

14.2

3 minutes to complete 2 rounds of:
10 overhead squats
10 chest-to-bar pull-ups
each set of 2 completed rounds, add 3 minutes and increase reps by 2

The women’s results are dominated by the ability to do a chest to bar pull up which stymied 9,179 women, or nearly 20% of the entire field making for a very short 10 rep WOD.  The men’s results reflect lots of “just lemme finish this set” pushes we see in the spikes at the right edge of each band.  The exceptions are reps 40, 88, 144, and 208.  180 athletes stopped at one of those scores even though they had just earned 3 more minutes to keep working.  Some days enough is enough.

14.3

 as many reps as possible in 8 minutes of:
10 deadlifts (95/135)
15 box jumps
15 deadlifts (
15 box jumps
…. increasing deadlift weight and reps(by 5) each round

crossfit_14_3_annotated

14.3 was the only WOD of the Open with increasing weights and the results reflect it.  The first 90 reps didn’t slow too many people down, but things got heavy in the 4th and 5th sets and the crowd fell off quickly. The elite athletes stack up nicely in a smooth curve depending on how much time they had left to hit those 205/315’s.

14.4

14 minute AMRAP
60-calorie row
50 toes-to-bars
40 wall-ball shots, 14/20 lb. to 9/10-foot target
30 cleans, 95/135 lb.
20 muscle-ups

crossfit_14_4_annotated

So, can you do a muscle-up?   22,052 (17%) athletes made it through the cleans but couldn’t get up on the rings, making it by far the biggest factor for both genders.  Toes-to-bar also proved problematic for 6% of women meaning 14.4 was just a quick spin on the rower.  You can see goal setting pretty clearly in the men’s cleans at 160 and 170.

14.5

21-18-15-12-9-6-3 reps for time of:
Thrusters (65/95)
Burpees

crossfit_14_5_annotated

note that lower scores are better on 14.5.

Finally, a nice clean distribution.  A WOD for time is a totally different animal. The median man finished in 18:36, just a minute ahead of the median woman at 19:28 meaning that 65/95 are pretty fair weights to compare the genders on thrusters.  We do see some spikes on round numbers, especially at the 30 minute mark.  87 athletes finished at exactly 30:00, but only 19 finished at 30:01.  That’s probably a combination of meeting a goal and wishful thinking when they scraped themselves off the mat and looked at the clock after that last burpee.

I got into this out of curiosity, but I hope this kind of analysis plays into the Open WOD programming from HQ.  Most of this years workouts probably did a good job differentiating between the regionals-caliber athletes because the tail at top end is generally nice and smooth.  The huge spikes in the meat of each distribution mean that a lot of the average athletes are getting lumped together or stuck on movements they can’t do with time to kill.

Let me know if you would like to take a look at the data or have suggestions for the next analysis.  Good luck to those athletes going to regionals, we can really see how many hurdles you cleared to get there.


– Sam Swift.  I’m a post-doc in the business school at Duke, a data-scientist, and 2 years into CrossFit at CrossFit Carrboro.  I’ve got more posts in the works, follow me @swiftsam.  Scraping and visualization code here on github.

Posted in Posts Tagged with: ,
  • Sean Burns

    Sam, I CrossFit with Kacy Tully and she shared the link with our box. I think your data is beautiful and shows that women need to focus on chest to bars and everyone needs to focus on muscle ups. I posted your link on the Outlaw Crossfit facebook page, a CF box in Alexandria with an owner/coach who is very analytical in his approach to CrossFit.

    • Sam Swift

      Thanks Sean, I hope they like it at Outlaw. I just got my 3rd and 4th ever muscle-ups yesterday before I posted this, so I’m ready for next year.

  • Nichole

    Will you do the filtering by height and/or weight? I’m interested to learn about the ideal body types for each WOD. I have a hypothesis about 14.5 — that the shorties had a good-sized advantage.

    • Sam Swift

      I do have all of the profile data too, but I haven’t looked at how it predicts scores in each WOD. Quick peak: there is evidence for your theory in 14.5, about 5 seconds / inch of height for women and 10 seconds / inch for men.

      • Kim Throneberry

        I’d love to see a follow up re: height/weight for the other WODs. There was quite a great deal of grumbling from “big guys” this year that most of the WODs were against them. I mean, my wall ball target is farther from my vertical reach, my pulls on the rower garner aren’t as long, my box jump is 33% of my height vs 30% or less, etc. Personally, I think it all balances out and more than likely people just want an excuse for why their weaknesses are what they are.

  • Elil Yuvarajan

    Sam, this is an awesome analysis! Very cool especially to see the spikes in the different WODs. I’d love to see a similar analysis for the athletes that are going to regionals by region. It’d be interesting to see if there were overall ways they approached the WODs.

  • Pingback: Competing In The Crossfit Open And What That Looks Like On Paper | Wild Rocket()

  • http://lc0.github.io/ Sergii Khomenko

    Hi Sam, nice analysis!

    Could you share your raw data? I have a couple ideas to check out with that dataset.

    • Sam Swift

      Thanks Sergii, I’ve replied by e-mail.

      • Keith Thompson

        I second Sergii, This is some fantastic analysis. I had guessed 14.1 would have some spikes like that, but I didn’t think they would be that substantial. Any way I could get a copy of your data as well? I’d really like to play around with them.

  • Pingback: Post Power Crops + Latest Jackets + CrossFit Open Side Note + More - Agent Athletica()

  • Pingback: The Monday Morning Roundup XVII: Crossfit safety and more – Vitality Sciences LLCVitality Sciences LLC()

  • Pingback: Tuesday, May 13, 2014 - Nashville, TN CrossFit -- CrossFit Forte()

  • datawod

    Sam, this is great. It inspired me to revive my blog and look at this year’s regionals, here: http://www.datawod.com/regionals-events-visualized.html. Would be glad for your thoughts. Cheers, Mike

    • Sam Swift

      Great post, Mike, I love how informative the simple histograms are. I was [/still am] hoping to find time to do more with crossfit data, but I’m glad you beat me to it with the regionals, and I love your presentation format.

  • Pingback: Crossfit Open 15.1: the data | Sam Swift()

  • Pingback: CrossFit Games Data 2012-2015 | Sam Swift()

  • Pingback: WOD Data: Crossfit Open 15.3 | Sam Swift()