I am not a number. I am a free man!

My tracking history

I have been tracking health, fitness and diet metrics since 2010. Back in those days it was all actual measurements and notebooks. Tracking food macros meant googling nutrition data one ingredient at a time and painstakingly summing it all up – which also meant restricting myself to a repetitive menu.

I started with weekly blood pressure readings and daily weight measurements – both of which needed to come down.

In the first heady days of a focused two year diet campaign, weight trended in the right direction and blood pressure did not get worse. As I hit the inevitable weight loss plateau, daily calories and macros (grams of protein, fat and carbohydrate) got added to the mix, and spreadsheets took over from notebooks. The early running watches with foot pods and heart monitor straps added more measurement opportunities, and as I got into more serious strength training, 1RMs (one rep max lifts) and body measurements joined the menagerie.

At one point, I was logging 47 parameters daily for nearly 18 months, covering nutrition, sleep, exercise, biomarkers, supplementation and some subjective attributes like “mood”.

That produced an impressive dataset that enabled me to explore correlations between variables. It was a big, time consuming project that demonstrated no significant correlation between supplements and any positive outcomes. I did however demonstrate a strong inverse correlation between Three Day Sleep Deficit and mood, strength gains and appetite control – in other words, I would have had better outcomes if I had spent more time sleeping, less time tracking and not wasted money on supplements.

While it is disappointing to know there is no magic bullet more potent than a good night’s sleep, this is still a valuable insight, and I continue to track a lot of things. Thankfully, the tools have come forward light years in the meantime.

What I have learned about tracking?

Be precise (or at least realistic)

We can measure length, weight and time reasonably accurately. Pretty much everything else is an approximation. This is a challenge if you are looking at small effects (e.g. standard advice to lose fat while strength training to gain muscle would be to maintain a mild daily calorie deficit ~400 calories – good luck with that if you cannot reliably measure calories in or out.)

It is pretty obvious that not every large egg has exactly 72 calories. Googling “food label accuracy” suggests that the labels may be off by as much as 20%, although I could not find an original source for that fact. Equally, when your smartwatch tells you that you burned X calories on that last set of bench press, or your biometric scale claims that you have Y% body fat, they are using models that can be wildly wrong – so don’t assume the numbers are accurate, and focus instead on the trends.

What I call Subjective Scale Metrics can also be problematic. Consider:

On a scale of 1 to 10

• How tired were you upon waking this morning?
• How sated were you after eating your breakfast?
• How hard was the run you just completed? (also know as Perceived Rate of Exertion or PRE)
• How happy did that donut make you?

These metrics can be really useful for testing the impact of experimental changes, but can you accurately and consistently distinguish between satiety of 3 or 4? I can’t. Where I do use subjective scales, I limit myself to range 1-5, and write a definition for each value. If it is too hard to write a definition, then the scale is unusable.

Beware the proxy metric

If you cannot come up with a reliable subjective scale, one strategy is to instead track something else that is measurable. For example I define Previous Night Sleep Deficit as 7.5 – number of hours slept. Of course this is imperfect as trackers do not absolutely accurately measure when you fall asleep and wake up, but they are reasonable, and I find this more effective than trying to say how tired I am.

Proxy metrics are everywhere. Body Mass Index (BMI) uses height and weight as a proxy for body fatness. It might give a reasonable indicator for an average member of the population, but with no accounting for strength and fitness, male athletes will tend to be classified as obese – the proxy model is too simple for edge cases. (If being a fit male is really an edge case?)

Other proxies just measure the wrong thing. Duolingo talk about their 2 million students with a streak of greater than 365 days. It is an impressive number and suggests that lots of people enjoy the app and are using it daily. But the real goal is to master a foreign language, not to maintain a streak – the proxy should be a model of the real goal, not a distracting alternative.

Sometimes you need a finer needle

Think about the old analogue gas meters; you could not tell if the gas was on somewhere in the house by watching the number dials; they moved too slowly. But there was that spinning wheel that reacted to any gas usage.

Weight lifting is measured in reps and weights. e.g. Bench Press 5 reps of 120KG. As the weights get heavier, I often hit a plateau where I cannot increase weight and keep the reps. Am I getting stronger from session to session or not? The measurement “perform 5x120KG bench press” is not sensitive enough to detect improvement from one workout to the next.

Recently I have augmented weight and reps with Velocity Based Training, using an app to track the speed of each rep. I can now see how fast I pressed the bar and how the speed deteriorates as I get tired over the set. This lets me track improvement at a finer level than just the weight pressed.

You can never measure everything

The point of health and fitness tracking ultimately, is to change or preserve some aspect of your health, and to learn what works (or not) for you.

You can measure a few things directly, and lots of things indirectly through proxies and reference models. But the human body is complicated and you are unlikely to measure (or even think of) every relevant variable. In the end, every breath you take and every move you make affects your body and most of it will go completely unnoticed!

In my 50s, I am far more aware of injuries and recovery times than I used to be. It would be helpful now if I had tracked dings and dents over the years to inform where I have potential weaknesses. Of course I cannot remember which shoulder I hurt 5 years ago on the monkey bars, or how long it took to heal – but there is a decent chance my body carries the result of my accumulated damage, and will punish me twice as hard for a repeat offence.

Then there are all the incidental movements that might not be inconsequential. I sometimes get soreness in my right knee, which I now attribute to always stepping out of heavy squats left foot first. I am testing that now by consciously alternating which foot I step out with.

Know why you are measuring

Tracking can be fun and sometimes even insightful, but it needs to be intentional; you need to be actually thinking about what you are seeing, and developing theories to explain the observations (i.e. apply the scientific method.)

Active Tracking (physically measuring and recording data points vs. passive data logging from fitness wearables) promotes introspection to find correlations. When you spot something interesting, you can run a mini experiment to confirm.

For example when I was regularly drinking a protein shake after lifting, I used to have it with milk until I noticed that I tended to lose bodyweight faster if I just used water. I was tracking weight and calories daily (if approximately) but the effect was greater than could be explained just by the calories in the milk. I tried a couple of weeks without milk and then a couple with. The effect was consistent and I no longer drink milk.

Some things I have learned through tracking

So having learned a bit about the art and science of tracking health, fitness and diet, what amazing insights have I uncovered?

I doubt any of it is truly amazing, but at least I have demonstrated to my own satisfaction that the following are true for my body in my 40s and 50s:

• A sleep deficit is bad for mood, motivation, training performance
• Milk inhibits weight loss more than the calories can explain
• I will lose body fat faster if I eat less than 150g carbs per day
• My resting heart rate is lower than average but will spike if I drink alcohol the night before
• Heart Rate Variability (HRV) has a reliable inverse correlation to stress, including overtraining
• 8000IU Vitamin D keeps my serum level at the target value – living in Germany and working in an office probably means I do not get enough sunlight despite trying
• Fast movements under load (kipping chin-ups, fast dumbbell rows, etc.) will trash my joints
• Tendon/Ligament pain takes ~6 months to recover and only if I stop repeating the damaging movements
• I will sleep like a rock after a hard squat workout
• Magnesium before bed noticeably reduces muscle soreness (and a double dose induces wild and vivid dreams)

Closing Thoughts

That seems like a short list of fairly obvious learnings – maybe I need to actively track my list of insights! There are always more tracking opportunities 🙂

Tracking can easily become obsessive and counterproductive. Just because you can measure something doesn’t mean you need to! While I enjoy my wearable tracking gear, a lot of what they passively log is of no real value to me – and if your want to get paranoid about dystopian futures where health insurance companies establish premiums and approve procedures based on ones internet fingerprint, maybe I actually do not want it all.

Tracking what actually matters, though, is really the key to changing/improving/maintaining over time and I do not expect to ever stop.

The important part is to recognise what actually matters, to pay attention to the data and the trends, and think about the root causes. Purposeful tracking is an active pastime and you will only get useful insights if you put the work in.

That said, our prehistoric ancestors were not spending their time counting their steps. They may have been counting days and tracking seasons and celestial movements, but otherwise, they were just getting on with it. I shouldn’t need a heart sensor to tell me I am stressed, or watch to tell me it is time to eat. The observable data is just a model and will always be complete. The real goal should be inner awareness to the point I can listen to my body and nourish it as it needs to be nourished.