Tuesday, April 10, 2018

Ignore the noise and look at only YOUR results

I'm happy to say that we've reached a point where the mainstream media and the trade magazines aren't totally scared about eating fat. I've been thinking about and reading about and experimenting with my diet for almost 10 years. I forget that not everyone has seen this and it even feels new (and scary) to many folks.

A member (I won't out him unless he says it's okay) sent me this article about fat for athletic performance and explicitly asked my thoughts about the MCT question. As you know I'm a big fan of MCTs (and yay to him for remembering I wrote about them 6 weeks ago!) for athletic performance (and other things too, like clarity of cognition). This article is about "busting the myths of fat" and is trying to tone down the excitement of using MCTs for endurance performance. I'll quote from them:
THE MYTH: Medium-chain triglycerides (MCT), best known for their presence in coconut oil, break down like carbohydrates, so you can eat them as an easy-to-digest energy source prior to exercise rather than traditional carbs. 
“However, nearly all of the research that exists on MCTs suggests that it does not improve endurance performance,” says Campbell. Recent research found that cyclists who used MCTs had worse sprint performance and reported gastrointestinal upset during intense exercise. In short: Keep it simple and stick to carbs before your workouts.
I don't doubt that there was research done to see if MCTs helped with endurance or intensity. But they don't quote what the study was looking for. Here's a little secret: even very highly funded nutritional interventions only last 4-6 weeks. It's just too expensive to feed people and keep them in the study longer that that. It often takes 8-12 weeks to become metabolically flexible to be able to actually use fat when you're accustomed to the standard American high carbohydrate diet. And yet, MCTs can cause Olestra-like experience if you eat too much too quickly - I even warned you about that. 

But rather than try to poke holes in the studies, I want to talk about why all studies are worthless.  I alluded to it in my last anti-industry post and the whole nonsense around the p-value.

Classical statisticians like to think in terms of rejecting or failing-to-reject the null hypothesis. What they can't do is estimate a probability distribution and see how the effects of something extend to a population. That is, they never frame their question as "how much increase/decrease in performance do we see in [insert sport here] from consuming a isocaloric MCT oil instead of a carbohydrate?"  They will pick a threshold of performance and test whether the MCT improves performance that much on average across all participants.

There are many things wrong with this, say 9 out of 10 people have a small improvement and the 10th has a massive decrease in performance. On average, that single participant could make the whole thing look like there's no benefit. Ideally, they would plot the probability distribution and you could see that lots of people had a small effect and one didn't. Then you could experiment on yourself and see how you perform - it's all about you, not the population, when you're trying to see what works.

My favorite applied math blogger (and yes, that's how sad I am, I have a favorite applied math blogger), had a great post about this last week. In his example, the question was "is the water safe to drink." The way you should go about it is test the water over time and space and plot the distribution of lead concentration. But what statisticians do is try to reject or fail-to-reject the hypothesis that the water is safe. It's unnecessarily convoluted, opaque, and wrong.

I understand why the p-value stuff was invented in the time before calculators when it was hard to estimate distributions. But in the world of computers and big data, we need to revamp how we do statistics and try to be more clear in our communications.

So, should you take MCTs to improve your performance? I don't know. The biochemistry suggests that there's something there. I do know that it works wonders for me, I feel better, move faster, recover better, and even think more clearly when I remember my MCTs before a workout. You won't know unless you test on yourself - the average of a noisy study will never tell you what your results are.  This is why there are thousands of studies that contradict each other - we're reacting to noise instead of the signal. People are messy and studies involving people are very noisy. We need to stop treating it as if science will give us the one true answer and start understanding probability.

I'm going to save this article and do a search-and-replace of MCT for "fat", "cholesterol", "HIIT", "coffee", "eggs", "meat", and any thing else that gets reported as science.

Oh, and keep sending me articles. I love it when you guys are engaged in fitness and science and want to share it with me. If nothing else, it gives me a blog topic.

Today's Workout - start lowering the intensity this week

press 5-3-3-3x3


deadlift 5x35%, 5x50%, 5x70%


light snatch day:
snatch, bell size -2, 10x(7R+7L)/1:00