The Calories and Weight Change Experiment – Results

A few months ago, I started the Calories and Weight Change Experiment. The idea was to eat at +500, -500, +1000, and -1000 calories from maintenance levels to determine the effect on weight change. Would my weight behave according to the classic, linear theory? Would it adhere to the set point theory? Would there be metabolic buffering instead?

The data were too noisy to say for sure with just me in there (let me know if you’d like to contribute your data), but they tell the basic story with weekly averaging of my daily measurements:

Red: +500 for 2 weeks. Blue: +0 for 1 week. Green: -500 for 4 weeks. Blue: +0 for 1 week. Orange: ended experiment, stopped tracking, and just ate whatever for 5 weeks.

I gained weight at about the linear rate expected while eating +500 calories per day, although it’s too noisy to tell. I then quickly returned to my starting weight after the first two weeks of -500. Makes sense, classically speaking. I then stayed there for two more weeks of -500 and the week of maintenance measuring. No further weight loss from my stable weight of 145.

It was really hard to eat +500 calories every day without resorting to more exercise, wheat, or junk food, and the only way I could do it anyway was to eat a lot of dairy. -500 was much easier, but after getting back to 145, it stopped having any effect. This is support for the set point theory. (Weak support, since there aren’t much data.) Given that the data weren’t very good, I decided to stop the experiment before killing myself trying to eat +1000 for two weeks without anticipating being able to learn much.

Then I just started eating whatever, whenever, and I lost another four pounds in five weeks (edit: now seven pounds in six weeks). I don’t know how much I was eating because I wasn’t tracking calories anymore. It was more than during -500, but better matched to my appetite. The only difference was that I got to stop eating dairy in order to meet caloric goals on those days where I was somehow under even on the -500. But yeah, this doesn’t support set point, or anything else.

The BodyMedia armband malfunctioned, the Zeo showed no change (and then started irritating my skin for the first time since I got it 1.5 years ago), the Quantified Mind experiment showed no relationship between calories and performance, my lifts stayed the same, and the bodyfat percentages from Withings and calipers showed variance within measurement error the whole time.

Not a very useful experiment.

Sorry, comments are closed for this post.