Strava's Fitness metric is deeply flawed

This Strava “fitness” metric is equivalent to Garmin’s “training load.” When you have peaked for a race, even after the race, your Garmin training load is decreased because of the taper. This is why your Strava “fitness” will go down after a PR race .

Simply put- Strava chose a poor name for this metric.

I'm not saying that anyone should care about their Strava fitness score, but it helps to understand where it comes from.

Strava Fitness = Longer term training load (recent weeks/months)

Strava Fatigue = Shorter term training load (recent days/weeks)

Strava Form = Fitness - Fatigue

AFAIK Strava's training load calculation is not published, but for runners who use HR monitors, it's going to be something resembling a weighted sum of your time spent in different HR zones. A minute near resting HR is worth ~nothing, while a minute near max HR is worth a lot. So, as you get more fit, the same performances will yield smaller training loads.

A few takeaways:

1) If your HR zones are set incorrectly on Strava, your training load is going to be wonky, which means all of this is going to be wonky. Also, factors that affect HR like altitude, heat, humidity, stress, etc. may affect your calculated training load.

2) "Strava Fitness" is not fitness, it's training load. Some people call it fitness, as the idea is that increasing your longer term training load will increase your fitness, but it's more precise to call it long term training load or long term stress or something.

3) A big Strava fitness score doesn't mean your fast, it means you spend a lot of time with an elevated HR.

4) These metrics are meant to help you manage training loads in a way that's more meaningful that just looking at miles per week. Just like miles per week though, you'll have to figure out what works for you (what's a good training load during a build up, what's a good form score in a taper, etc.) but when you find that, the hope is that you can make your training more productive/less counterproductive.

Thanks,

by chance are you employed by Strava?

I have other feedback I can give privately

Raphael Warnock Knock wrote:by chance are you employed by Strava?

No, but I've spent some time in the bike/tri world, where these metrics are more widely used.

knox harrington wrote:

1) If your HR zones are set incorrectly on Strava, your training load is going to be wonky, which means all of this is going to be wonky. Also, factors that affect HR like altitude, heat, humidity, stress, etc. may affect your calculated training load.

Also, if your HR monitor is wonky, all of this is going to be wonky. Many people measure HR with an optical sensor on their watch. This is a very convenient way to measure HR, but unfortunately not a very reliable one. YMMV and some people have no issues, but if the watch is too loose, or too tight, or too close to your hand, or your skin is the wrong color, you may have issues with it.

Jon Arne Glomsrud wrote:

silviuburceadev wrote:

Couldn't care less about Strava estimations. Runalyze is spot on on predictors for me.

Funny you mention it, but Runalyze also uses training impact (TRIMP) based on HR and they use the same line of thinking around fatigue and fitness. Maybe presented differently and with slightly different formulas, but basically the same idea.

TRIMP is along the lines of Garmin's EPOC Training Load. However, computation of the latter uses FirstBeat's heart rate variability algorithms; Garmin's heart rate variability is not something Strava (nor Runanalyze?) would have access to.

Jon Arne Glomsrud wrote:

silviuburceadev wrote:

Couldn't care less about Strava estimations. Runalyze is spot on on predictors for me.

Funny you mention it, but Runalyze also uses training impact (TRIMP) based on HR and they use the same line of thinking around fatigue and fitness. Maybe presented differently and with slightly different formulas, but basically the same idea.

It is, indeed, but there are other metrics that Runalyze uses, e.g. marathon shape, which is a new concept compared to Strava & co.

seth_da_runner wrote:

Strava is just a social media for runners and a subpar running log

Why is it a subpar running log? I use it more or less only as a log (without the social media aspects), and it works fine for me, certainly better than anything else I've tried.

It's junk and you should have an option to turn it off. My fitness was apparently at its lowest for three months when I ran a one minute PB for the 10k.

The metric also places too much weight on distance rather than intensity. Just running a load of junk miles at moderate pace will get you a great fitness score but I doubt it will translate into better performances

maybe it's like golf where a lower score is actually better. strava has punked all of us with the fitness score lol.

that said, i've never looked at mine.

On a slightly different topic, the relative effort score seems to give a somewhat reliable reading as to how much a workout actually took out of you.

IMO it still overvalued total time and undervalues intensity. I’d also be curious to see if there is anything built into the algorithm for training modality. If I putter around on my bike for 90 minutes, I get a much higher relative effort score than if I go out and run hard tempo for 20 minutes, although the tempo run leaves the legs much more fatigued

It's worse than that. My mile race that left me totally drained registered as "easier than your usual effort."

If I do a long run it will say "this was harder than your usual efforts"but if I do a 5k pr it will say "this was easier than your usual effort"... seems to only take distance and time into account.

seth_da_runner wrote:

Honestly, no one should be measuring their fitness based off of what Strava says, nor would I recommend using Garmin's VO2 max. Strava is just a social media for runners and a subpar running log, it's fun to use but shouldn't be much more than that.

+1, this is an accurate statement.

Strava has more bugs than any other apps. But I use it till now because it gives me a pleasant user experience. Anyway, five months ago, I recorded my first run here and my first workout at the gym. So I think I will not give up this application soon because I have a more pleasant experience with it than not. But, I have to tell you that it's hard to do sport when you don't do more than 20 years. But, work on self-discipline is worth it. I see the result; I feel better. I try to read as much as possible about nutrition. I have discovered exciting articles about food and sport. So I read daily to get informed.

https://www.fitnessbaddies.com/cheap-adjustable-dumbbells/

ScotchRunSleep wrote:

This Strava “fitness” metric is equivalent to Garmin’s “training load.” When you have peaked for a race, even after the race, your Garmin training load is decreased because of the taper. This is why your Strava “fitness” will go down after a PR race .

Simply put- Strava chose a poor name for this metric.

I kind of agree, but understand that they chose the name. Fitness does also relate to the accumulated load. In theory. The more intensity in training, the more load, which directly translate to higher fitness. But fitness does not equal performance. And using the freshness is important. The more you "rest" the higher freshness and in the moment some time after a high fitness value when the freshness increases, there is a crossing point or a point where you peak.

Fitness is based on fatigue which is based on load, and it is based on heart rate. My HR monitor died during vacation so no HR and my fitness went down like I did no training and the freshness got up like I was always resting. HR is a good indicator for aerobic workouts, but less for intervalls and speedwork, which I think is very important for performance. So it is limited in being a good indicator.

I fell into a habit of trying to maximize fitness so I saw to that the average HR was as high as possible in the workouts. That is a bad idea, I tell you... I have stopped using fitness as a training indicator. I used Elevate app on top of Strava (former Stravistix). It was nice, but based on the wrong indicators for a well made training plan to run fast, I think.

As near as I can tell, the Strava metrics are deeply flawed for runs where your HR varies over the course of a run. For example, I suspect it would rate the first run as much harder than the second.

1) 20 minute tempo at HR 160.

2) 10 minute warmup up at HR 120. 20 minute tempo at HR 160. 10 minute cool down at HR 120.

Why? Because it averages about the HR and thinks you didn’t work hard on your second run.

Said differently, it seems to all be about the average HR and a period of easy running kills the average

hr stuff wrote:

As near as I can tell, the Strava metrics are deeply flawed for runs where your HR varies over the course of a run. For example, I suspect it would rate the first run as much harder than the second.

1) 20 minute tempo at HR 160.

2) 10 minute warmup up at HR 120. 20 minute tempo at HR 160. 10 minute cool down at HR 120.

Why? Because it averages about the HR and thinks you didn’t work hard on your second run.

Said differently, it seems to all be about the average HR and a period of easy running kills the average

I think you are wrong, it is not this simple. Take a look at their explanation. Of course they do not make stupid things, but there are limitations to how useful the metric is.

From Strava:

"Fitness

While fitness is a complicated concept, it can be simplified to an accumulation of training. The Fitness Score is calculated using Training Load and/or Relative Effort to measure your daily training, and an impulse-response model to quantify its effect over time. This will intuitively capture the development of fitness from training, as well as the loss of fitness during a break."

And:

"How It's Calculated

Our method for calculating Fitness, Fatigue, and Form is based off an impulse-response model first developed by Dr. Eric W. Banister in 1975. It was later applied to cycling by Dr. Andy Coggan.

The concepts apply to any measure of training stress. For example, the first models used average heart rate and time. We use Training Load and Relative Effort, computed using data collected from a power meter and heart rate monitor, respectively."