This discussion has been locked.

You can no longer post new replies to this discussion. If you have a question you can start a new discussion

VO2Max overestimated

Former Member over 8 years ago

http://www.runnersworld.com/sweat-science/can-your-watch-estimate-your-vo2-max?utm_source=t.co&utm_medium=social&utm_campaign=sharebutton

As commented in this forum, based on this study the Garmin VO2Max is overestimated.
It is interesting to see the hr strap is still more accurate than the inbuilt hr for VO2Max results.
Now we know why those predicted times sometimes seem unrealistic.

0 WillNorthYork over 8 years ago

I think the online VO2-to-race predictor calculators are based on Jack Daniels' research. This is one of the calculators I use, which happens to produce numbers similar to Garmin.

http://www.runningforfitness.org/calc/racepaces/equiv

Ok.. I’ve looked at the table and did a little additional research on the matter…First off, this table was obtained from research Jack Daniels did for thousand of runners that he has trained…He has tested thousand of runners and performed a correlation between their running times and their VO2 max…the table represents an average..you can find the table in his marathon books (Daniels Running Formula). Second, the VO2 results and predicted running times are an average..As such, some people running times will be better than predicted and some will be worse.

Third, I’m not definitely sure, but I believe most of the runners that tested were running 50-100 miles per week..As such, if you are not running anywhere near this mileage, this prediction table is probably useless.

Many of the comments on that page repeat the common refrain that their predicted race times from VO2 max are wildly optimistic. And there's the common answer that these are idealized times which you get with "optimal training" under "ideal conditions".

I think it's a similar idea to race predictor calculators (even McMillan) which take a race time for one distance and predict your other races. They are supposed to give you results which are accurate assuming you train ideally. (In my experience these calculators are pretty accurate for distances below full marathon)

This is probably the formula that Garmin uses, which afaik are derived from empirical studies on a wide range of runners. I haven't read Jack Daniels, but maybe I should.

(I think Garmin uses this formula, or something similar, because the calculator closely matches my results from Garmin at least -- my VO2Max has varied by 8 units over the last year, so I've had a couple of data points.)

http://www.runningforfitness.org/faq/vo2-max

How do you estimate VO2 max from race results?
The formula used in the calculation is taken from Daniels & Gilbert (see
references):

VO2 Max=(-4.60 + 0.182258 * velocity + 0.000104 * velocity^2)/(0.8 + 0.1894393
* e^(-0.012778 * time) + 0.2989558 * e^(-0.1932605 * time))

Note that velocity is in metres per minute; and time is in minutes.

edit: I checked against a couple of results from the vo2max accuracy thread. They're close but not exactly the same; in some cases, closer than others.
https://forums.garmin.com/showthread.php?66329-Vo2max-accuracy

Maybe garmin is using a formula derived from newer research?

0 Former Member over 8 years ago

Great Info

Great Info everyone, I appreciate all the links and detailed information.

Regarding modifying the max HR when I am in the cold location to keep consistent the vomax number in the watch, I think that would be like to cheat the algorithm. In fact, I think the garmin in this case is doing a very good job, because I think the vomax is the measure of the performance of your aerobic engine. In my case my engine gets affected by the temperature and humidity and that is what the garmin is telling me. Perhaps and instead of cheating the algorithm, it should have as an inputs temperature and humidity variables which are constant in the lab when doing the vo2max test.

Again the exact number may not be useful for me but the trend is what is important.

(again after 2 runs in Florida back from Buenos Aires, my Vo2max is now 2 points down and my training status became unproductive! Garmin connect says: "Your training load is at a good level, but your fitness is decreasing. Your body may be struggling to recover, so be sure to pay attention to your overall health including stress, nutrition and rest") This use case scenario is so tricky and of course I wouldn't expect the watch to figure it out.

0 runningfan97 over 8 years ago

I think the online VO2-to-race predictor calculators are based on Jack Daniels' research.

And that is where the problem starts: Daniel's tables are based on VDOT and not on VO2max.

0 WillNorthYork over 8 years ago

And that is where the problem starts: Daniel's tables are based on VDOT and not on VO2max.

Yeah, I have heard people make that point. Jack Daniels refers to it as a "pseudo vo2 max".

http://www.runnersworld.com/workouts/threshold-training-finding-your-vdot

An aerobic profile involves identifying a vVO2 Max (velocity at VO2 Max) that represents the speed of running a race that lasts about 10 to 12 minutes. This vVO2 Max reflects the runner’s economy and VO2 Max and will be the same for all individuals of equal race ability—although one runner might accomplish his or her vVO2 Max with great economy and a relatively meager VO2 Max and another runner with not-so-great economy and a high VO2 Max. It doesn’t matter how the components vary if they combine to provide the same result. Basically, Jimmy Gilbert and I placed every runner of equal performance ability onto a common economy curve, which meant they would also have the same mathematically generated VO2 Max and a similar lactate-response curve. Equally performing runners are assigned equal aerobic profiles, which means they would also have an identical pseudo VO2 Max but not necessarily the VO2 Max they would show in a laboratory test.

Instead of referring to this pseudo VO2 Max (the one based strictly on performance) as VO2 Max, we use the term "VDOT." VO2 Max is properly stated "V-dot-O2Max."

But there's no other way to estimate race performances from VO2Max (or vice-versa) (in the absence of other parameters), afaik, except to pretend VO2Max and VDOT are equivalent.

Whether Garmin should be trying to estimate race performance solely from VO2Max is debatable, I guess.

0 runningfan97 over 8 years ago

But there's no other way to estimate race performances from VO2Max (or vice-versa) (in the absence of other parameters), afaik, except to pretend VO2Max and VDOT are equivalent.

No, absolutely not. As VDOT and VO2max are seldom the same, one can not treat them equivalent. Adding some sanity checks would definitely help with the estimation as Garmin has a whole lot of data available. LTHR and race results for instance.
As mentioned in an other thread, I have a very high VO2max on Garmins (F3 and 935) which is currently at 61. My current VDOT is more in the range of 52 or so.
If you look at runalyze, they manage to have quite good VO2max estimations based on your running data. Still a little high in my case, but 54 to 52 is a whole lot closer than 61 to 52.

0 WillNorthYork over 8 years ago

Thanks, I'll def check it out! Looks very interesting.

I meant in the absence of other data/parameters, they're treated as equivalent; e.g. for the purposes of these online calculators. I agree that Garmin potentially has a lot of other data (of course), but one problem is that it's not always available. An FR230 won't have LTHR data. Not every runner will have race results. I'm sure it's easier for Garmin to just do a simple lookup based on treating their estimated VO2Max as VDOT and call it a day.

0 WillNorthYork over 8 years ago

Wow, awesome site! Thanks! Ran across it searches before, but never tried it out until now,

So I uploaded 2 races from last year when I was in good shape.

For those 2 races, runalyze tells me:
- My VO2Max (by HR) is 1 or 2 units higher than Garmin's prediction (which leads to wildly overestimated race results). My time by HR is also wildly faster than what I actually ran (which makes sense)
- My VO2Max and VO2Max (shape) are 10-11 units lower than the above. And if I plug that into an online VO2 to race predictor, I get similar results to my actual race results (which makes sense)

If I'm understanding this correctly, runalyze is telling me that I could've run much faster had I maxed out my average HR during those races. (To whatever's physically possible for that distance, not my actual max HR obv.). Also suggests to me that Garmin's predicted VO2Max/race results are based on an idealized maximal effort.

The two races were:
1/2 marathon at 80% Max HR
30k at 83% Max HR

Took a quick look at a letsrun thread with runners much faster than me: their HRavg for a marathon was higher (87-89% of HRmax)

Kinda fits my intuition that I have been held back by lack of strength and/or injuries. When I got in better shape I rarely maxed out my HR during workouts (maybe once every 1-2 months). Or maybe my max HR is just set way too high. But I can still hit that number "whenever I want" (as recently as 2 weeks ago, in normal temperatures); I just don't want to because of injuries/lack of strength.

(I should also mention that I got the same numbers in runalyze before and after I set my max HR on the site.)

0 runningfan97 over 8 years ago

Walpole provins

Yes, runalyze is a great tool, absolutely. Two runs though are definitely not enough to get good predictions and analysis. They calculate some correction factors over your different workouts and have powerful options to adjust the calculations (e.g. when HR spiked, temperature was too high, too hilly courses, training races, etc.). Over time, they come quite close to what your current fitness status is.

And what the guys at LR wrote is true, maxing out HR in races is something, one got to learn through a lot of races. With not so experienced runners normally muscles and/or mind first give up before letting one go in the red (or when it comes to marathon: orange) zone.

The discussion shows that there is interest in the analysis of the data. Guess that's one reason why we running tech nerds buy these expensive watches :). A lot of improvement can be made though helping in the analysis and interpretation of the data. I am confident that AI and Big Data mechanisms can be of good help in the future. Currently they only do a table lookup for race predictions :cool:

0 WillNorthYork over 8 years ago

Yeah...I am in process of exporting everything from GC >_> using a handy python script runalyze suggested. First there was a problem with special chars the script didn't like (emdash), now there's a problem with some invisible half-deleted activity in GC. #technology

Definitely looking forward to crunching the numbers! Amazing that runalyze is free so far, but I'm sure they'll get monetized/bought out in the future.

It's also amazing what can be done for free outside of the GC ecosystem, but we're willing to pay for in the name of convenience. (Like training load; anyone can get TrIMP if they sync their activities to strava and use Chrome...)

0 runningfan97 over 8 years ago

Yeah...I am in process of exporting everything from GC >_> using a handy python script runalyze suggested. First there was a problem with special chars the script didn't like (emdash), now there's a problem with some invisible half-deleted activity in GC. #technology

Definitely looking forward to crunching the numbers! Amazing that runalyze is free so far, but I'm sure they'll get monetized/bought out in the future.

It's also amazing what can be done for free outside of the GC ecosystem, but we're willing to pay for in the name of convenience. (Like training load; anyone can get TrIMP if they sync their activities to strava and use Chrome...)

Good luck with the mass import. I did that end of last year and moved away from Sporttracks (the desktop version) and GC to runalyze. Wasn't that easy as my data was ranging back to 2012 when I started running with Garmin, before that I only had Excel sheets and estimated data.
It wasn't that easy to be honest, especially I wanted to take over additional data which is not available in the FIT files (suppose, you are using the same python script as I did). For me it was crucial to have the titles and notes from the workouts also in runalyze. So I wrote a Java program to support the fit file import. After that, there will still some discrepancies where the exported GC was strange and I had to manually change these. Now I have a full set of data ranging back to 2006 and am perfectly happy.

I know the guys from runalyze, they are also nerdy runners and are doing this in their sparetime. They are happy for every donation they get.