I just got a 920xt and it provides me with a VO2 estimate after each workout. Last night was a 65 after a longer Zwift race, which seems a little optimistic for a 58 year old man. I’m good at making 4 wkg with a hr under 135, but my top end is nothing special. Years ago I did a real test and it involved riding to failure.
I’m flattered, but sceptical.
Garmin estimates won’t be as accurate as numbers you’d get from a lab test. For some people they will be spot on, for others they’ll be quite away from their real values.
Also, Garmin updates your vo2 after each workout, whereas it actually does not change that often.
4w/kg ftp is in the 60 ml/kg/min vo2max range. 65 ml/kg/min is closer to 4.5 w/kg.
The algorithm is sensitive to heart rate and the MaxHR setting in your heart rate zones. So, that needs to be accurate.
So, if you do a qualifying workout where you have a low relative heart rate, and higher power then you will get a higher vo2max score.
Its been discussed here many time before. The heart rate basis makes it sensitive to all the same things that can influence heart rate day to day.
Is “gain” in the title meant to read “Garmin”?
I find the VO2max values provided by my 920XT do track well against my perception of the trends in my ability and comparison between current and past ability. So I think changes in the value do offer a useful measure. However, I can’t say how accurately the value provided is likely to match a lab test derived VO2max result. I also don’t know all the inputs Garmin use to generate that value. Obviously weight and power(bike)/pace(run) data are primary factors but I suspect heart rate and other data may also be taken into account. If so, the accuracy of the data you provide may be critical. So, are you sure you’ve got accurate weight and max heart rate values entered, for example?..
I agree that the trend is more indicative than the number. But, I’ve never been tested. However, my race results align closely with the garmin and other online calculations.
According to the firstbeat white paper: age, weight and hrmax. Also according to the paper the method has an 8% error for cycling, and 5% for running. The sensitivity to errors in hrmax is just under ~1/2% per bpm additional error in the estimate, outside a 5 bpm window. It is less sensitive to over estimated hrmax than under estimated.
Longer sessions are more likely to provide an accurate measure. 75% at 2k (running), 95% at 6k.
https://www.firstbeat.com/en/aerobic-fitness-level-vo₂max-estimation-firstbeat-white-paper-2/
However, my race results align closely with the garmin and other online calculations.
Based on my experience, it actually used to be quite optimistic. Now with the most recent version of the race predictor (FR945, F6) I’m a bit faster than the watch’s estimates but it’s really close.
My Garmin gives me the same estimate for both running and cycling. 55. I’ve always wondered how close it really is. From my bodyweight and race performances (mostly running), it actually seems pretty accurate.
In order to get the VO2 estimate, the system compare your power (for bike, or speed for run) to your measured HR.
But your HR zones must be set correctly before that.
And you can only set them correctly if you have been tested before, and know your :
FTP power and corresponding HR (bottom of zone 4)
VO2 power and corresponding HR (high of zone 5)
and ideally also you z2 levels.
If you do that correctly, it is pretty accurate (but subject to usual HR variations induced by heat, bike cadence, digestion, condition, …)
There’s no mention in their paper of zones at all. Strictly hrmax. The method extrapolates submaximal power /pace to hrmax. Honestly, most of the secret sauce seems to be in assessing the “reliability” of the data, and excluding unreliable segments from use.
The algorithm is sensitive to heart rate and the MaxHR setting in your heart rate zones. So, that needs to be accurate.
So, if you do a qualifying workout where you have a low relative heart rate, and higher power then you will get a higher vo2max score.
Its been discussed here many time before. The heart rate basis makes it sensitive to all the same things that can influence heart rate day to day.
On that basis too, your Garmin VO2 will fluctuate based on whether you are doing more activities of a sport you are stronger in than a sport you are weaker in. My Garmin VO2 plummets in the winter because my Zwift cycling activities don’t report/affect the Garmin VO2 but my runs still do and, um, as most cyclists turned triathlete… ugh running… Spring is always a fun time though, VO2 gains a couple points every week for awhile 
I agree with your “secret sauce” point, using only the “valid” sectors.
I agree HR max is a key point.
I do not know if their algorithm take care of the intermediate zones (in order to separate 2 guys with 180 HRmax, one with rest at 45bpm, the other at 80bpm), the only thing I now is that, in my case, putting HR and Power zones correctly (from real lab test), it is now working very well, giving realistic values in line with lab test (lower when undertrained, aligned when on shape).
I suppose these intermediate values help, but I can’t prove it, as I did not try to put stupid zones values.
Is “gain” in the title meant to read “Garmin”?
I find the VO2max values provided by my 920XT do track well against my perception of the trends in my ability and comparison between current and past ability. So I think changes in the value do offer a useful measure. However, I can’t say how accurately the value provided is likely to match a lab test derived VO2max result. I also don’t know all the inputs Garmin use to generate that value. Obviously weight and power(bike)/pace(run) data are primary factors but I suspect heart rate and other data may also be taken into account. If so, the accuracy of the data you provide may be critical. So, are you sure you’ve got accurate weight and max heart rate values entered, for example?..
No, it’s garmim 
I just got a 920xt and it provides me with a VO2 estimate after each workout. Last night was a 65 after a longer Zwift race, which seems a little optimistic for a 58 year old man. I’m good at making 4 wkg with a hr under 135, but my top end is nothing special. Years ago I did a real test and it involved riding to failure.
I’m flattered, but sceptical.
I had a VO2max test done at Henry Ford Hospital and the Garmin VO2max gives me nearly the same number for running.
Hope I am still putting out 4 wkg for an hour at 58.
Hope I am still putting out 4 wkg for an hour at 58.
That has stuck withe over time, it’s the shorter efforts that have deteriorated with age
.
It seems that Zwift data uploaded to Garmin doesn’t affect your Vo2 score, but if you capture it on a Garmin device to begin with, it will.
Recently bought a FR945, until I connected my turbo, HR and cadence as sensors I was only getting “credit” towards Training Load Focus (must be purely based on HR), and afterwards is when I started to get metrics for Vo2 Max / Aerobic / Anaerobic.
It’s a pain recording things twice, so now I only populate Connect with data collected on a Garmin device - my Zwift rides/runs go to Strava only.
According to DCR, workouts on a Forerunner, Fenix or Edge will update/sync via Connect, but not from Zwift data uploaded from an iPad (in my case anyway)
Would love to get to 4w/kg at any point in my life!
The vo2max calculation is only done in real time by the garmin device, not by post processing on garmin connect.
I’m in the same boat, I end up recording on the Garmin but then the Zwift gets loaded twice to strava. I just have to clean it up after.
I like this 920, I was going to sell it and go back to my 310 but it’s growing on me,especially the battery life
I was run tested at a facility last Feb. I just upgraded to a garmin that has vo2max estimate last week. The estimate seems pretty close for me. It was a point or two lower than the Feb test on my runs this week, but close enough as I wasn’t running max and figure the estimate algorithm is truly only an estimate.
I was run tested at a facility last Feb. I just upgraded to a garmin that has vo2max estimate last week. The estimate seems pretty close for me. It was a point or two lower than the Feb test on my runs this week, but close enough as I wasn’t running max and figure the estimate algorithm is truly only an estimate.
That would be neat if I was at 65 for my age, that seems like a decent value. I think Zwift racing has been worthwhile this time of year