The Problem with Fitness Scores

Greg Marshall
2 days ago
4 min read

Is your training app running the show?

Runner looking at mountains with fitness metrics overlayed

“All models are wrong. Some are useful.”

-George Box

It used to be that we decided for ourselves how we were doing. I still remember my old manual training log where I’d enter my mileage for the day and a quick thought about how it went.

But the roles have reversed. If training apps and sports watches are the new training logs, then it seems more like the training log is telling us how we’re doing instead.

There’s a common favorite way to do this. It’s the big number in the middle of the screen whenever you open the app: the fitness score. It looks like it makes things easy. If the numbers go up, you must be fitter. If the number goes down, better get moving.

Great in theory, but is that reality?

Stress is not adaptation

What training apps commonly package as “fitness” is more accurately called stress. There are other questions to be asked about the reliability of stress tracking, but that is their aim. You can look at any platform’s definition or formula for how their “fitness” score is calculated, and what it boils down to is how hard the algorithm thinks today’s workout was, compared to what the algorithm believes is your current capability.

There’s a problem, though. The body gets fitter from adaptation to stress, not the stress itself. Meaning, you can experience a stress that is followed by lots, little, or no adaptation at all. In fact, it’s one of the main reasons why it’s harder for older adults to maintain muscle mass like their younger selves; the same stress yields less adaptation in the aging body. But in the app, the fitness number responds the same way, whether you adapt or not.

How fast is fitness lost?

There is also the problem of detraining. According to fitness score equations, if you run an effortful workout one day, and rest the next, the fitness score will take a hit because of the rest day. Meanwhile, timelines for physiological adaptation vary greatly across body systems. If your body is still undergoing the chemical changes that improve fitness, why is the score telling you that your recovery is decreasing fitness?

Actual research on detraining tells a different story. In athletes who followed an extended period of training with a period of reduced or absent training, VO2 max typically fell by a little bit less than 4% in the 30-days after cessation of a training routine– versus roughly a 50% drop in fitness score that would occur over the same 30-day timeframe. The study also found that partial cessation of training and total cessation didn’t show a substantial difference in the rate of detraining. Regardless, no studies are assessing detraining within about 72 hours of a training bout, because reduction of physiological measures within that window are typically attributed to fatigue or exercise-induced damage, not due to detraining effects.

Metabolic and Mechanical Stress

There’s enough evidence available to develop better paradigms than popular fitness metrics are conveying, but there’s a third problem. If fitness scores aim to measure stress, the question should be asked: can stress even be effectively captured by a single number?

Exercise routines don’t stress the body in one way, but in many, widely varying ways. In running, we have a metabolic stress each time we head out on an hour-long jog. Metabolic stress is the chemical reactions that occur to liberate energy. Yet, we also have the mechanical stress for that same run, meaning the contraction of muscle fibers to generate force, the micro-trauma to bones as they abruptly impact the ground with each step, and so on.

Consider an example: two athletes experience the same metabolic stress (same total calorie burn) on a hilly two-hour run. The context is that one athlete typically runs up and down steep mountain trails each day as part of their normal training. The other athlete usually trains on flat, paved city sidewalks.

Before and after the run, the athletes have matching fitness scores in a popular training app.

Those two athletes are likely to experience wildly different stresses in their legs from the same hilly long run, to an extent that one of them might develop an injury while the other feels unfazed the next day.

What it all means

Tracking stress over time is useful. Coaches, athletes, and apps should be trying to capture it. We can learn from it, and subsequently improve our exercise routines. The problem is that fitness scores don’t capture what they advertise (fitness), and they also don’t capture stress very well.

The tools for better stress tracking exist. Dynamometers and inertial measurement units (IMU) hold a lot of promise for revolutionizing the way we train and track stress. The embedded systems company XSens produces an IMU suit that allows researchers and clinicians to better assess the forces each body segment is experiencing. It might be a great solution for training athletes too, if it can be offered affordably. Unfortunately, XSens currently comes with an eight thousand dollar price tag.

In the meantime, the rest of us might be well served to at least partially revert to the old days, when we told the tracker how we were feeling, rather than letting the tracker tell us.

Don’t stress over your fitness score.