Garmin Body Battery slammed indirectly by Altini: “Made Up Scores”
In his regular newsletter, world sports-HRV expert, Marco Altini, had a no-holds-barred tear into many of the ‘Made Up Metrics’ used by today’s crop of wearable tech. This includes Garmin Body Battery, an industry favourite similar to those from many other brands.
Altini’s article follows my critique of Siepe et al.’s study, which found that “Garmin Vivosmart can’t measure stress,” and how that study was reported in mainstream media outlets like The Guardian and Wareable.
Garmin Vivosmart can’t measure stress: Study Shows Little Correlation to Self-Reported Levels
Here’s what Altini has to say, and I have to agree with what he says almost word for word. My only criticism was that he used “Made Up Scores” whereas I prefer “Made Up Metrics” :-).
Here’s what the Marco Altini newsletter article adds in case you don’t subscribe:
- Explicit Condemnation of Continuous Stress Monitoring: Altini stated that the Siepe study “put the application of HRV-based continuous stress monitoring as proposed by wearables to the test, and found that the correlation between wearable data and self-reported stress scores was ‘basically zero’“. He emphatically declares that “there is no such thing as continuous stress monitoring using HRV” and calls these applications “a wild extrapolation merely serving user engagement”. I’m perhaps sometimes sensationalist on this site, but Altini’s view is a stronger, more categorical denouncement of continuous stress monitoring than mine.
- Personal Stance and Prior Work: Altini additionally states, “Continuous stress monitoring was the nail in the coffin for me when it comes to wearables, as the misuse of HRV data for this application was something I couldn’t get behind“. He also explicitly refers to his more detailed discussions on this topic, which I summarise, edit and link to in an article he wrote on this site.
- Broader Critique of “Made-Up Scores”: Beyond just stress, Altini extends his criticism to other “made-up scores” in wearables, such as “readiness, recovery, etc.”. He asserts that “none of these parameters are good at tracking what they claim to be tracking”. Again, this site and one or two others have quietly said this for some time.
- Distinction with Appropriately Measured HRV: Altini highlights a recent study that found a negative correlation between self-reported stress measures and HRV when measured during the night and a positive correlation with resting heart rate. He states that “night data provides a good assessment of an individual’s resting physiology and stress response”, emphasising that HRV can be a sensitive marker of stress when measured correctly. This contrasts with the flawed continuous monitoring approach discussed in the Siepe study.
- Direct Advice to Users: I concur with Altini’s advice to readers – “Keep that in mind before you invest your mental energies and trigger additional worries and concerns for you or your athletes because of made-up data with no physiological meaning”.
Take Out
I could have one of those ‘I told you so’ moments, but I will refrain as it’s Monday 🙂 And I know many regular readers are aware of this in any case.
There you have it. These all-day stress metrics, like Body Battery, are nonsense! Albeit good fun.
The story gets worse. I am in some private discussion groups with various people in the industry. I/we can assure you that all major wearables companies have excellent in-house HRV science advisors. The companies all know precisely what they’re doing and the flaws of their various approaches. The information on this site and from Altini is nothing new to any of them.
I believe commercial pressures are being applied to deliberately misstate the usefulness of many of the current crop of wellness metrics…aka Made Up Metrics.
Last Updated on 27 January 2026 by the5krunner
Reader-Powered Content
This content is not sponsored. It’s mostly me behind the labour of love, which is this site, and I appreciate everyone who supports it.
Support the site: Follow (free, fewer ads) · Subscribe (paid, ad-free) · Buy Me A Coffee ❤️
All articles are written by real people, fact-checked, and verified for originality. See the Editorial Policy. FTC: Affiliate Disclosure — some links pay commission. As an Amazon Associate, I earn from qualifying purchases.
tfk is the founder and author of the5krunner, an independent endurance sports technology publication. With 20 years of hands-on testing of GPS watches and wearables, and competing in triathlons at an international age-group level, tfk provides in-depth expert analysis of fitness technology for serious athletes and endurance sport competitors.
Very interesting
I agree that Body Battery and even things like whoop strain is nonsense to an extent. However, I do find value with Garmin stress measurements in very specific circumstances. It can give a useful indication of how I am recovering from training stimulus when I am seated and at rest (low stress = higher hrv = body in recovery mode vs. high stress = lower hrv = body not recovering well). Also interesting to see how stress spikes a bit after eating, before gradually going back to baseline levels.
it’s interesting but not science based.
whoop strain is not nonsense i think its actually pretty good but am fully aware of others disagreeing (was only chatting with @fellrnr last week on that subject as it happens), maybe whoop recovery is (edit: nonsense) in the sense of the above article
How is whoop so much better then Garmin is they are both using the same type of data as input?
not sure which metric you are referring to . i was also unclear – see clarification in previous comment
i agree whoop’s recovery metric maybe nonsense in the context of the article.
there was a reasonably well reported scientific paper in the last few weeks which looking at a nightly average hrv reading vs an orthostatic reading. i think it was linked to acclimatisation. basically the nightly average reading measures resting physiology and didn’t work, the orthostatic one did (paraphrasing). the latter being more of a metric liked to sports physiology
WHOOP Strain is a measure of cardiovascular and muscular exertion that quantifies the amount of physical and mental stress you’re putting on your body.
https://www.whoop.com/us/en/thelocker/how-does-whoop-strain-work-101/
This sounds the same as all day stress metrics
ah good point.
i think the muscular bit was added later. the core of the reading i’m reasonably sure was originally based on eTRIMP.
I believe whoop also uses hr and hrv in particular to determine strain score, in the same way that Garmin uses it to determine body battery.
And to be clear — body battery itself is not a useful feature for me. Stress measurements at specific times (e.g. when sat on the sofa) is what I find useful, and I’d argue that garmin’s data is fine in this case (wrist-based hrv is much more accurate when not moving, typing, etc)
Agree. The study is against self reported stress levels. Self-reported.
We’ve seen how valuable self reported information is in VAERS.
I think this is a self indulgent article for clicks and likes.
Perhaps instead of making up a straw man (measuring perceived stress), they should actually find our how Garmin define stress and measure against that.
Do better.
it’s self reported but it’s also comaring that against something that isn’t a valid comparison.
the authors did say they couldn’t get the garmin published algo.it would be nice to know i agree
Garmin just avoids killing me every other day by capping the body battery to 5
Lmao I’ll sleep great and still wake up with 65 body battery, and by noon my body battery is at 20. Waste of time metric. The sleep tracking and heart rate are all I care about.
Lmao I’ll sleep great and still wake up with 65 body battery, and by noon my body battery is at 20. Waste of time metric. The sleep tracking and heart rate are all I care about.
And I sleep terrible, wake 5 times and by morning have a Garmin body battery of 95.
///“made-up scores” in wearables, such as “readiness, recovery, etc.”. He asserts///
Just to clarify, this is the same Altini that took a job at Oura in 2020 as a ‘Data Science Advisor’ per both his site and theirs (and both sites currently list him as such)? And then to further clarify, the same Oura that has:
– Stress Score
– Readiness Score
– Sleep Score
– Resilience Score
– Heart Health/Age Score
– Cardio Capacity Score
Or, is that a different Altini? Just want to ensure I’m aware of all the Altini’s out there.
He is also the one behind using hrv to measure stress:
https://www.hrv4biofeedback.com/
and the more popular hrv4training
same one.
Lots of good people work for these companies, presumably they give good advice about what science supports and what it doesn’t. someone else probably makes the commercial call.
as per the article above and Altini’s article, all the wearable companies have these kinds of metrics not just Garmin. Garmin only came up in the context of the (flawed) scientific study referred to
Of course, Garmin has invented many algorithms and statistics. However, there’s a method to this madness. As a dietitian and active athlete, I feel that subjectively, the vast majority (70-80%) of my results align with my well-being. But for me, both in work, sports, and life, it’s the trend that counts, not specific numbers. And Garmin is simply right. When HRV or Body Balance is abysmal, I don’t feel well either. When stress is higher during a given period, Garmin immediately gives me a message that I’ve had a difficult and stressful period. Furthermore, from a physiological perspective, there’s a concept of oxidative stress that we can’t detect, which accumulates, among other things, through too intense or too frequent training. And no research with subjective interviews will change this, because in such cases, a person’s opinion is irrelevant. A damaged central nervous system reveals itself over time… Altini… doesn’t comment on the guest because he’s not an outsider, he works for a sports equipment company, so his opinion doesn’t count. It’s distorted and biased. Amen.
HRV can be higher when you are ill and it can be higher after hard exercise.
even production of ROS from exercise is not so striaghtforward, exercise can increase antioxidant capacity. i dont know anything about measuring it but if you could measure it, the absolute level would not tell you anything in itself (I need to know more about this topic i think, i’ve meant to look at it for a long time)
“Lots of good people work for these companies, presumably they give good advice about what science supports and what it doesn’t. someone else probably makes the commercial call.”
My point is that this isn’t a case of a random employee, it’s a partnership with financial incentives. Both sides entered into the partnership agreement for publicity and credibility purposes. After all, they don’t tend to publish other employees names publicly. Oura wanted his name at that point in terms of credibility, and he wanted whatever gains came along with it from his side. And both sides still do.
It’s really hard to understand when one is taking the money/gains from a company partnership/sponsorship, yet concurrently saying the product doesn’t work. Which, is literally what he’s saying. Why continue it? And why continue to promote it on both sides?
Ok i see your point.
i dont think he’s saying the product “doesn’t work”, it’s a less binary perspective than that. he’s definitely saying the overnight average readings trended over time mean something when there are deviations, from memory he also says oura’s product is one of the more accurate and recommends it in his app (don’t know if that’s commercially driven or not)
loved the Youtube vid on NOMAD, btw. nice work. i added a link back to it.
As you already know it is the same one. Here is the Pubmed link for the work he published in collaboration with Oura.
https://pubmed.ncbi.nlm.nih.gov/34201861/
I do like that they have to show conflicts of interest by stating that the writers are both employed by Oura.
This is the other interesting side of the argument. There are so few peer reviewed studies on wearables, but many of the ones that are done are done by the people making the products. (just like so many other parts of our corporate science community)
I do wonder how the accuracy of the HRC data impacts the metrics. No one talks about what the polling rate of the sensor is so hard to know how precise any r-r timing data is. If you follow any of the talk over alpha 1 dfa there is lots of talk over how important the data coming in is both clean of artifacts and precise so can’t be measured with optical sensors. The metrics from wearables may not be as sensitive and only work at lower HR values so don’t need the high precision as alpha 1 during exercise but better data in should lead to more useful metrics
Sure people look into the HR accuracy of the wearables but not HRV (I acknowledge that its very hard to check HRV accuracy)
check out my hrm600 review, i correlated the hrv to two other sources…it was wrong but trended correctly.
the accuracy of the hrv will affect some of the metrics a lot. even with a 2minute chest strap resting reading in the morning i can take a second reading immediately after and get quite a different result SOMETIMES
Very. good point! I think some wearables smooth out readings rather than tell you their findings were incomplete or not clean.
I bet it uses the same random number generator algorithm as the gradient field. And the weather widget.
some of the gradients readings can change when stationary! i think i wrote about my experience testing a couple of the edges. it must use the difference in 3d gps points to get the gradient on some occasions – hence not actually moving but the imagined gps point drifting is bad!
It started happening on my 1040 solar after the firmware « merge » with the 1050 – Before that it was pretty accurate. My ancient edge 800, despite lagging by 20-30 meters is spot on. Leave it to Garmin to fix a bug and introduce two new ones, then reintroduce the original bug several revisions later and come full circle.
@Paul, not sure about that.
I’m pretty certain that the 1040 had incorrect gradients soon after launch. I even remember one particulet incident on a rather precarious and steep off road hill, or ‘flat’ as garmin called it
I think stress recording has been very useful for me, especially when recovering for illness, COVID or food poisoning as an example. It may be high orange all day whilst I’m ill and I can see it reduce as soon as I’m starting to feel better.
I’m an IT worker who sits down for a living.
When I spend the weekend doing construction on my house, my Garmin reports high stress, and my body battery goes way low.
If I do a very intense run, the stress stays high for 6 hours.
To me that’s feels like the Garmin is reflective of the reality of my physiological state.
So how can the body battery and stress score be utter garbage like this article is claiming?
When the guy claims that the wearable bears no relationship to how stressed the users feel, is he thinking stress means ‘worried about something ‘ or do the wearable users realise stress means physiological stress?
I think the article is too polarised too be useful/meaningful.
Since Oura was mentioned above…
I am recovering from a COVID infection (positive test 9 days ago). I wear a Garmin 965 24/7 as well as an Oura ring.
Oura’s metrics were the only one to pick up on anything being possibly wrong during these last 9 days when I had a period of “minor-to-major strain” for 3 days. Bear in mind I have been pretty careful in my recovery and making sure I have gotten adequate sleep etc.
Garmin noticed nothing and kept telling me I was “Well Recovered” in my Training Readiness and Body Battery showed absolutely nothing amiss. HRV also “balanced” during this time as well.
I love my 965 for active training and as a Smartwatch for notifications/payment etc but the Body tracking is absolutely garbage.
It is interesting that we live in an age where science is no longer science and if we really really believe in something (especially marketing hype that has been packaged really well and pounded in for years) then science or research must be wrong if it contradicts what we think is true. Add in a dash of tribalism and you have a great “Say what!, that’s bullsh$t!” cocktail. A “I know what I know and I don’t need your study” sandwich.
Most of the people reading this article (both the original from Marco Altini and the one here at the 5k Runner) are probably reasonably in shape. Most of us bounce back really really well and even though some are desk jockey weekend/evening warriors, our bodies are really really good at bouncing back and going for another round. But when you add in a little drama to show how hard you worked in a graphic, man that makes the journey even more meaningful.
Garmin offers some amazing tools and I believe the mean well with every new metric they add, however there is an issue when there is no peer review. Even with COROS and their ECG that is not certified, that is probably not a tool I would rely on without a peer review. I don’t think that Guardian’s study or Altini’s work is trying to shame anyone or make people feel bad. If anything it is coming from a place where they/he are trying to save us money and help us concentrate on what is really useful and meaningful. They want to help us clear away some of the fluff and get to real sciency bits.
When claims approximate the appearance of science closely enough, they are often granted credibility without scrutiny.
I forgot to say thank you to the 5k Runner for having blog posts like this. It is a rare find in this day and age where influencers are battling for corporate dollars. I love that you present different sides to arguments and thoughts,
ty
Well said. Reality is what’s local to the observer, and it doesn’t matter if N=1, or if it’s contradicted by science. This applies to many things.
For my N=1, Garmin’s HRV score can be confirmatory, but not predictive. Body battery, performance condition, etc are useless and can be demotivators. In a broader context I find these metrics to be largely a sales tool. I also cringe every time I see someone that doesn’t know if the heart rate their watch is showing them is correct. And people will believe in some magical metric over what they feel – I wonder if we’ve lost touch with our own bodies.
Garmin body battery is an extremely useful predictive (rather than lagging) indicator of energy levels for those suffering from ME or long COVID.
I don’t believe completely in those stats but I always think they have some meaningful side.
A few years ago we did a bike ride with friends. I didn’t sleep well that night, only a few hours of sleep, thus I started the ride with a body battery of only 43. I rode around 310 km’s that day, making the body battery of course drop to zero. Sleep section after that continued to record stress instead of resting. 🙂 The next day stress bars were high to the top and body battery was still zero. Also I was tired as hell and even had hard time sitting. 🙂 The following day my body started recovering from the ride and my body battery started to increase. It was 66 in the morning and could recover to 75 at 12am and 100 after few more hours of sleeping.
So I know those stres bars aren’t always about how stressed I feel but how stressed my body. Also sometimes they seem to know what elephants doing in my head better than me. Body battery is kinda different, it can’t go deeper than 0. But I wish it did. If it could go deeper than that, we’s definitely see how bad our lifes are. 🙂
This is what I did to check my body battery and stress levels with my Garmin Fenix 7x:
Beta blockers (5 mg): Slowed down my heart rate, which led to higher HRV. While doing regular daily activities – desk work, weekend shopping, and errands – stress levels stayed well-aligned. I felt restful, and stress remained very low even when I was active.
Beet juice + pumpkin/sunflower seeds (for nitric oxide & magnesium): No drastic changes, but I noticed a moderate reduction in stress and heart rate.
Adjusting coffee, tea, and high-glucose foods: Stress levels increased significantly when glucose spiked after meals. Elevated stress lasted 3 – 4 hours until glucose returned to normal, meaning that frequent meals kept stress levels high most of the day, with recovery only happening overnight.
At the end of the day, these devices aren’t directly measuring stress, HRV, or even heart rate – they’re predicting them. The approach seems to be: ‘we have more sensors now, so let’s use them for probability-based predictions rather focusing on improving accuracy. That’s why, with every new release, the numbers keep shifting. Apple disrupted the market by introducing far better predictive algorithms, bringing their smartwatch readings closer to the accuracy of dedicated devices like the Polar H10, pushed Garmin to do similar predictions over improving the sensor accuracy.
as you say, apple Watch’s ohr sensor is quite accurate for some things
but its HRV readings are taken about once every 15 minutes, on average, if you are still enough.
Could that be the reason why Apple don’t give us those metrics? At the moment, many small app developers try to copy bodybattery, readiness score and so on…..
possibly. i did think that too.
however Apple gives sleep stage stuff which will also be “subject to error”
i find the body battery the most useful metric of them all. please prove it is not a ) a real metric
b ) does not show what it should have
c ) propose alternative in the market
thank you
Without touching scientific reasoning, just staying at the untrustable level of an individual’s impression:
– I removed Body Battery, Training Readiness and Sleep Score from the glances/widget I check periodically;
– and I kept HRV status.
My observation was that my Enduro 2 was the first Garmin which could at least identify the total sleep time properly, but the score allocated was not in line with personal feelings.
On the other hand all night HRV score was in line with that.
Do not ask me how it camn be it is just how I felt in the long run.
On the other hand I could easily challenge Body Battery score showing that there were another hidden and uncharged Reserve Body Battery.
Training Readiness was a close call it is the one from the three removed data which might be added again.
Btw I read their definitions and the articles about them, so even I am surprised that why HRV status was approved by me, but not Body Battery.
Garmin’s HRV tracking has been useful for me.
It’s very difficult to distinguish between a common cold, body aches, and influenza (which can be detected with a PCR test at the hospital).
Last winter, I woke up with a cold. My HRV graph was so horrible (flat…, healthy, the graph constantly fluctuates). I went straight to the hospital for a PCR test and was diagnosed with influenza B.
This is crucial.
If you take antiviral medications for influenza within 48 hours of infection (the sooner the better), recovery is significantly faster.
I received the antiviral medication Peramiflu injection 24 hours before I was infected, and I was completely cured the next morning.
And then I ran a 10k immediately.
The elephant in the room is software patents. No watch vendor publishes directly peer reviewed papers on the methodology, hardware interfaces, and algorithms. It ends up to be magic numbers because they don’t want to expose how they do it and at the same time they don’t want to risk lawsuits on potential infringement. As a result, instead, you get these funded studies by third parties that attempt to correlate the features with user experience. Sad trombone…
yes, somethimg like that
i think the formula has to be published for it to be patented, the patent filing would then expose the ‘secret’ and the competition could then do something similar provided there is a different method.. a case in point being the current apple spo2 case where using the same feature but in a conencted way seems, according to apple, that they are not then infringing the og patent.
so. keep it secret.
I guess there is then always the risk of it being ‘black boxed’. AI must make that process easier to discover the formula
but if the formulas are just invented out of thin air (a lot of them are) then why bother to try to replicate it in the first place? simply invent your own magic number.