Tuesday, April 5, 2011

An ideal CrossFit weight (for men)?

Ooooo, good stuff today.   All the athletes have submitted scores for week 2, which means tons more data.  Logistically, I should be able to make plots a little quicker because I shouldn't have to reacquire all the personal data (weight, height, region) from everybody again.

The crossfit open doesn't use weight classes, and the argument put forth has been something along the lines of, "The workouts have all been balanced so heavy weight (easier for heavy people) are balanced with body weight exercises (easier for lighter folks).  Is that true?   Let's look at some data.

The blue line represents the 'elite' or the top 10% of athletes for a given weight (5 lb increments).  The red line is the overall average athlete for a given weight.


Plotted is the performance of athletes across weights for workouts 1 and 2.  To explain further - I have grouped athletes according to their weight (in 5 lb increments), and plotted the mean performance of those athletes.  'Average' athletes in a given weight are plotted in red, and 'elite' althletes, or the top ten percent of athletes in a given weight, are plotted in blue.  For reference, I have also included the cumulative percentile score chart on the right, since that information isn't immediately obvious on the open website.  If you wanted to quickly see where a score of 200 ranked on workout 1, you would go directly up from the bottom axis at 200 until it crossed the 'S' shaped curve.  From that point, going directly left to the axis will give the percentile score, which in this case is around the 30th percentile.

For workout 1, the curves go up from a weight of 140, peak at 185, then head downward.  Does this make a huge difference?  I think for average athletes the answer is yes.  An 'average' 185 open athlete bested 20 percent more people than his 'average' comrade weighing in at 230 lbs, translating to an overall rank difference of over 2000 people!

What about for elite athletes?  The mean percentile scores for 140, 185, and 230 pound elites were all in the top 10% of total scores, so I think weight wasn't too much of an issue for those folks.  The overall pattern is the same though.

As an aside, why should we care about the 'elite' catagory of athlete?   Given that there's 17 total regions, and 50 athletes that move onto the regionals (total 850), I would argue that the top 10% of athletes (~1100) should stand a reasonable chance at moving on to the next round of competition.  Understanding patterns in these athletes should hopefully tell us in the future what's important for the open.


It's good be a 180 pounder so far!


The circumstances change a bit when you look at workout 2.  Everybody, say over 220 pounds, should have screamed a collective Bender favorite, "We're boned!"   Yikes the heavier people were punished on this exercise,  no matter if you were an average athlete or an elite one.   In the elite category, the difference between an average athlete at 180 lbs and one at 230 was a crazy 19 percentile points.  Another stat, no person weighing above 220 lbs cracked the top 5% of scores.   I don't know how that will translate into a real effect at the end of the open, but if this trend continues I wouldn't be surprised if there weren't many 220 pounders that make it to Regionals.

So to answer the question, in there an ideal CrossFit weight for men?  So far in the open, the athletes around 180 pounds have had it pretty good.  There's still 4 workouts to go, so circumstances could change...

Analysis for the women will hopefully come next!

20 comments:

  1. This is great analysis. Keep it up. It will be very interesting to see what the ideal weight is after all 6 WODs.

    ReplyDelete
  2. Hey J, sweet site! I'll be following this the entire time. Is there anyway to look at the standard deviations around the mean at various weight classes? That might be an interesting analysis, because my guess is the performance variance is likely higher around the edges of the graph.

    Kyle, a 140#, 10% dude.

    ReplyDelete
  3. Workout 3 just made the little guys gulp and the big boys a bit happier...

    ReplyDelete
  4. @Dale - I just thought the same thing! I'll be lucky to hit the 1 mark on this one.

    @zlooop - I haven't looked at that specifically, but it's an easy plot! One modification though, I think we would want to plot the CV for each weight class, which normalizes the spread around each mean.

    ReplyDelete
  5. I wonder if this data would if this data would hold up to the masters class. I imagine that it would, but would love to see the number for us elders. The third workout is a little bit much for masters. Just hoping not to DQ.

    ReplyDelete
  6. Would be interesting to see height graphed like this too. I'm thinking 5'10-6' at 185 is the sweet spot.

    ReplyDelete
  7. Very interesting. In addition to weight, I wonder what other variables matter.

    ReplyDelete
  8. This is awesome! Keep up the good analysis! This is beautiful stuff!

    ReplyDelete
  9. Second the comment from (unreadable code) above.. really curious to know how height correlates to this data.
    Great work!

    ReplyDelete
  10. Very interesting analysis, and nice presentation of the data - thanks. I am in the Master's category and am a bit under 140 - I think this workout will clean out some of the lighter folks, particularly in the two younger Masters divisions. It would be interesting to see an analysis at the end of which weights get dnf'd along the way. I tend to think the heavies can survive, if not prosper, throughout, but that the lighter weights will get significantly trimmed. Good incentive to get stronger!
    - Steve Meis

    ReplyDelete
  11. Great analysis...keep it coming! Out of curiosity...How are you getting the raw data? Are you mining the site?

    ReplyDelete
  12. Just a thought,

    Instead of arbitrarily picking the top 10% as the cutoff for elite athletes,
    fit a smooth curve to your percentile chart data and choose a significance value for the derivative of the fitted curve...then drop a line from the intersection of the derivative @ the point of significance to the raw performance axis and use that value to delimit your elite athletes for graphing performance vs weight

    ...some non-elite athletes might make it to regionals, but don't have a chance in hell of making it to the finals...would be neat to see the size of said pool...

    ReplyDelete
  13. Very cool.
    Can you add in the "ideal age" stat for a crossfiter?

    ReplyDelete
  14. This is awesome! Plz do the same for the ladies!

    ReplyDelete
  15. I think a performance vs weight/height (ie BMI as crude as it is) would be a more accurate determinant of performance.

    ReplyDelete
  16. I think you're missing an opportunity by not also taking into account the height of the athlete. 185lbs at 5'5" isn't the same as 6'3" particularly for overhead lifts. Why not create a weight:height ratio and graph that?

    ReplyDelete
  17. This is great stuff. It confirms quite a bit of what I was already thinking too. Thanks.

    ReplyDelete
  18. I would like to see if the ideal weight changes after further workouts get posted. WOD 11.3 and 11.4 were much more strength based and I wouldn't be surprised to see the trend move up a bit.

    ReplyDelete
  19. good share n completd and easy to undrstand

    ideal weight

    ReplyDelete
  20. WOD 11.3 and 11.4 were much more strength based and I wouldn't be surprised to see the trend move up a bit.asdad Where Can You Buy Gynexin Online?

    ReplyDelete