First, the let's take a look at a few stats about the programming. As I noted on Twitter a few weeks ago, this year's programming appeared to be heavier than last season, but pretty typical compared to 2011-2013. After factoring in the actual loads on the snatch event, here's what we saw for the load-based emphasis on lifting (LBEL) in 2015, compared to prior years:
- 2015 - 0.69 men, 0.46 women (48% lifting)
- 2014 - 0.59 men, 0.37 women (43% lifting)
- 2013 - 0.60 men, 0.38 women (43% lifting)
- 2012 - 0.92 men, 0.60 women (67% lifting)
- 2011 - 0.68 men, 0.44 women (48% lifting)
- Average 2011-2015 - 0.69 men, 0.45 women (48% lifting)
As we can see, 2015 was very average as far as loading for the Regionals. There was really nothing outlandish in terms of required loads, and limiting the snatch to two attempts kept the weights lower in that event. In fact, the men's average lift was 232 lbs., and when we look at the men who were in the top 330 in the Open last season (roughly those who would have qualified in the Super Regional format), the average snatch was 235 lbs. And keep in mind that last season was a hang squat snatch, rather than a full snatch, any-style. The women's load did go up from 145 lbs. to 150 lbs., which is probably a testament to the continued improvement of lifting skill among female CrossFit athletes. The limited number of attempts also increased the variability compared to last season: the standard deviation in 2015 was 51 lbs. for men and 32 lbs. for women, whereas in 2014 it was 33 lbs. for men and 28 lbs. for women (again, limiting to the top 330 from the Open).
Another question is, well, does the programming really have an impact on who qualifies for the Games. The answer, in my opinion, is yes, although the effect is rather small. The chart below shows the average weight of the men's Games qualifiers (ignoring top 2 and bottom 2 values each year), as well as the LBEL of the programming at Regionals each year. Weights for 2015 based on data from Sam Swift, prior years were collected manually from the Games site when I wrote last year's Regional Review.
The effect is relatively small, but the weight of the qualifiers does follow the same pattern as the LBEL of the programming. In 2014, when the LBEL was at its lowest, the average male qualifier was 192 lbs. In 2012, when LBEL was at its highest, the average male qualifier was 199 lbs.
Additionally, I looked at returning regional athletes to see if there was any correlation between weight and the change in rank from year to year. The hypothesis was that since the programming appeared to be heavier this year, bigger athletes should tend to see more improvement in their ranking than smaller athletes. Again, the effect was not huge, but it was consistent with expectations. For men, there was a -12% correlation between weight and change in rank (negative change in rank = improvement), and for women, there was a -8% correlation.
Another thing we saw with the programming this year was an even bigger emphasis on Olympic lifting, with 48% of the points coming from Olympic-style barbell lifts (36% coming from the snatch and clean alone). There was also a decline in basic gymnastic movements, such as the toes-to-bar and pull-up, while high-skill gymnastic movements, such as the muscle-up and handstand push-up, continued to be a major focus. Basic gymnastics made up only 10% of the points, compared with about 20% in 2013-2014, while high skill gymnastics made up 33% of the points. High skill gymnastics made up 36% of the points last season, but only about 15% from 2011-2013.
Combined, Olympic-style barbell lifts and high skill gymnastics made up 81% of the points at the 2015 Regionals. That is the highest of any HQ competition in history. Below is a list of the competitions with the highest percentage of points coming from these two sets of movements:
1. 2015 Regionals - 81%
2. 2014 Regionals - 67%
3. 2015 Open - 61%
4t. 2010 Games - 50%
4t. 2011 Open - 50%
To me, we had a little bit too much emphasis in these areas. These are the two most technical types of movements, so I'm not sure if these are necessarily the same athletes who will excel at the "unknown and unknowable" events at the Games. At Regionals this year, there were also no powerlifting-style barbell lifts, no kettlebell or dumbbell movements and no wall balls. There were also no burpees, just like the Open. The last individual HQ competition prior to 2015 without burpees was the 2007 Games. That just seems odd.
All this being said, I liked the programming this year more than last season. The programming was balanced as far as loading, and the events were pretty well-designed and competitive. Sure, I'd have liked more than 2 attempts on the snatch, and I think event 2 ("Tommy V") was a little boring for the fans, but generally I thought the programming made sense and was good for the fans. Event 7 was another thriller, although I'm still a little bit more of a fan of last year's event 7 (pull-ups/OHS).
Finally, let's look at the qualifying athletes. Of the 40 men's qualifiers, only 15 were first-time athletes. Two athletes, Spencer Hendel and Nick Urankar, returned after missing the Games for at least two years (Urankar had missed three years). Of the 40 women's qualifiers, only 13 were first-time qualifiers. Thuridur Erla Helgadottir also returned to the Games after missing the last two years.
How did my predictions turn out? Actually not bad. I'd like to have a few back, like Stacie Tovar and Lindy Barber each at 4% (probably should have cut them a little more slack for just having a bad year in 2014), but overall, things were pretty well-calibrated. The mean-square error (MSE) was 7.35% this year, which is worse than last year's 4.0%, but things were also tougher to predict this year. Because of the smaller field, a higher portion of the field had a legitimate shot to qualify, which is generally going to make the MSE higher. Had we just given every athlete an equal shot to qualify, the MSE would have been 10.7%, whereas last year it would have been 6.1%.
The chart below shows the calibration of the model this year. You can see that across the whole field, the actual predictions (blue line) were rarely off from the perfect predictions (red line) by more than 5-10%. For instance, the far right blue dot indicates than I gave the best athletes about a 73% chance of qualifying (x-axis), and in reality, those athletes qualified about 68% of the time (y-axis). Not bad!
Before I go, I want to mention the CFG Analysis Games Pick 'Em, which we'll be doing again this year. Be on the lookout for all the rules to come out in the next 3-4 weeks, along with my predictions, which you will be gambling against. I'm planning to make a few improvements, and I think it should make for another good time. If nothing else, it gives you a few more athletes to root for in Carson. See you in a few weeks!