Modelling the bodybuilding process

One of the biggest challenges in starting the process of bodybuilding is looking in the mirror every day. You see the person you are in the mirror, weighed by the baggage of the person you've been seeing for years - not the person you're intending to become.

Visualisation is key to motivation. To help with that, I've been working on some linear models to predict the effects of my workouts, plotting ahead towards the 9/1 deadline I've set for myself for Burning Man 2015.

To start with, my local gym, Fitness SF, has an InBody 370 - which does basic measurements of total body water, dry lean mass, lean body mass, body fat mass, weight, skeletal muscle mass, etc. To the quantified self fanatic, this is a lot of data - it's a pretty good breakdown of your entire body composition, as measured through the electrical impedance across multiple electrodes, held at hand and foot. Of course, there's error - confidence intervals aren't shown, nor is there any clear idea of just how much the margin of error really is; and the process takes long enough that I'm not about to do it every day. It is, however, enough to show continuous progress if measured, say, every 2-3 weeks.

So the process begins by taking the data from when I started measuring - February, roughly three months after I started my workout regime - until 'now'.

Base linear models (GLM, Quasi family) for total body water, dry lean mass, and body fat mass used in downstream models.

Without any data on limitations of my body, it makes some pretty hefty presumptions - that effectively, I'm just going to keep growing until I look like Hulk Hogan... or The Hulk. None of which is true - my frame isn't going to support an infinite amount of muscle, there's a natural ceiling long before that moment. Projected forward infinitely, I'm just going to keep adding weight until my neck cords snap from the weight.

To add some realistic restrictions to the models, we can turn to research on pro bodybuilders. WeighTrainer has an e-book that reviews some research, and after reviewing data and building models based on the data of pro bodybuilders, came up with an equation for predicting maximum muscular bodyweight. Paired with a prediction on percent body fat, it can compute a 'maximum' for your frame based on three additional variables - your height, your wrist circumference, and your ankle circumference. Now, this data isn't perfect - in fact, while it is reasonably predictive of the data he analyzed (about 300 class and overall title winning drug-free bodybuilders and strength athletes from 1947 to 2010), it's likely to be a wildly optimistic prediction.

It is, however, enough to add limits to the models that bring them down from extreme projections; for each datapoint, I can compute what the 'maximum' lean body mass and weight I could potentially gain are, based on that measurement's body fat rating, and then compute the same for the forecast data based on the body fat forecast. That lets me build models that include the relationship between weight, muscle mass, and body fat as they relate not just to each other at the time of measurement, but to an 'optimal' limit state.


At my current body fat percentage - call it 24 - these models suggest I have a 'maximum' lean body mass of 207 lbs, with body weight at that percentage at a hefty 280 lbs.

Projected further out, these models don't yet 'bottom out' properly - but they curve hard as the linear projections for the base metrics reach the target max potential: around 199 lbs of lean body mass at 15% body fat, and a total body weight of 235 lbs.

Recomputed lean body mass, weight, and skeletal muscle mass predictions incorporating the limitations of my physical frame

This brings the models into a much more realistic alignment, without needing to apply arbitrary weighting or constants, which do a poorer job of backtesting against historical data.

This is far from good enough - but it's a start, and something to keep me motivated to push hard as I exit bulk phase in August.

These models suggest that I'll add another 5 pounds of lean body mass, and put on another 2 pounds of actual body weight in the process (shifting fat into muscle). The other thing this shows you is that my caloric intake isn't high enough - I shouldn't be burning fat like this while I'm bulking, but I'm eating as much as I can without literally making myself physically sick from it.

I'll have a lean body mass of 177 lbs, and weigh roughly 223 lbs, at about 20% body fat as I enter cut phase. If I do exceptionally well, I might be able to get that down to 17% - which, while leaving me with lean mass far (30 lbs) from my maximum potential, would still be an astounding improvement to what began this process in November as a stick figure in a Google tee shirt.

I can be happy with that.