MinnMan

(yes I know this is mostly a zombie thread. I didn't want to start a new thread, though....)

Does anybody have an inkling as to why I'm getting such wildly different numbers on Strava for energy (kJ) and calories here? The two should be close, so I'm guessing that Strava is getting the numbers from different places (i.e., taking the calories directly from Wahoo, calculating the work by its own algorithm, or vice versa)?

This is a winter ride, so no power meter on that bike, it's just HR data, recorded on my Wahoo Element and transferred to Strava.

This mismatch is typical of what I get on my winter rig. On the road bike with the power meter, the values are in much better congruence.

(I know calorie burn from HR data has low accuracy - I'm not asking that question. The work calculation should have similarly low accuracy, but in any consistent calculation, they should closely match.)

WhyFi

I'm not seeing two sets of numbers to compare, but one of the things that I've noticed with Strava's power/kj/calorie figures is that, when no power data is available, the type of bike selected will have a big, big impact on their estimates. One of my club mates would go for a road ride on a bike labeled as a gravel bike (and it is a gravel bike, but with skinny slicks) and Strava would give power estimates that were out of the realm of possibility for him - he's strong, but not nearly as strong as they were giving him credit for.

MinnMan

The "Energy Output" is given as 1251 kJ, and the "Calories" is given as 2635 kCal. With standard conversions (4.184 J/cal and 25% work efficiency), these numbers should be very close to one another.

It's true that estimates based in part on "rider weight" may have extra uncertainties when one is riding a heavy bike and wearing 10 lbs. of outerwear, but that still doesn't account for why energy output and calories burned are so out of whack with one another.

WhyFi

Oh, okay - I thought that you were talking about Garmin figures vs Strava figures. Yeah, that's a pretty big disparity. I would have assumed that they'd use the better estimate of the two (based on available data) to estimate the other, but I don't know what the hell is going on with this. I guess they're estimating both separately, calories based on HR and work based on bike type and assumptions on associated resistances. Do you think that that 95w average is low? I wonder how changing your bike type and then refreshing the activity would impact things.

Iride01 

Calories to me are always dietary Calories and never should be used for power. Power if one has a power meter, will give the best estimate of Dietary Calories used for any one particular ride.

The Calorie estimates given by the device or any site is based on things that might at times be totally out of whack. Maybe when it's available, sites should use PM data for calculating Calories, but I doubt any do and maybe there are some reasons why they don't.

Essentially it's apples and oranges. And everyone has different varieties of each.

oris

This thread blew my mind a bit and ran into this article from Trainer Road while Googling it more. It explains a bit more in depth from MinnMan's response.

https://www.trainerroad.com/blog/cal...201%20Calorie.

MinnMan

Yes, that's the crux of the question. To highlight the quote in your link

Iride1 's post is besides the point. I thought I made clear that I wasn't asking about the absolute accuracy of the calories counted (or the energy expended, or for that matter, average power). I'm wondering why the energy expended and calories are so discrepant, when most calculations bind them quite tightly.

fooferdoggie 

my bosch powered E bike has a watts meter but I use a garmin or such to record my rides to strava. usually they are about 50 calories off on my 9 mile commute. the garmin only has my heart rate the bike has my heart and watts. but when garmin uploads to strava the watts are way low. it has the garmin calories. I can compare them over and over.

genejockey 

Yeah, your Calories seem out of whack. I just checked my Sunday ride, which is in many respects similar to yours - HR average and max within a couple beats, mileage within 2 - but with 2000 feet more climbing and a higher average speed. But higher I got a higher estimated Kj and over 1000 LOWER estimated Calories.

MinnMan

Thanks for those data points. I agree that of the two numbers I'm getting for my rides, it's likely the calories that aren't as accurate and the energy is likely closer. This statement is based on my perceive effort vs. calories and energy for road bike rides with power meter.

Elsewhere it's been noted that Strava simply imports the calculated calories from the calculation done by the head unit (can anybody verify?), but I don't see what settings (rider weight, etc.) could be so far off on my head unit. And if that's the case, from where does Strava get the energy calculation? So I'm puzzled.

fooferdoggie 

with a garmin it takes the calories garmin calculates. with my karoo2 it does not calculate calories unless you have a power meter then it goes more just by speed same as what I get when I did not wear a HRM.

Iride01 

I don't think Calories are a straight calculation based on only heart rate. Back years ago when there was a lot of discussion on the Garmin forums about Calories being off, the general consensus from those that seemed to have a good idea about it was that Garmin looked not only at your actual HR, but how fast it ramped up or declined compared to your speed, body weight and other things.

So if you allow for that and the way a program would have to be written to deal with that, then there is a lot of room for the number to come out way different from one same ride to the next. Might even in some cases just be a unfound bug in the program that isn't worth the effort to look for.

MinnMan

This is how Wahoo calculates calories from HR data
evaluating this equation once per second



But if not from HR/calorie, how does Strava calculate energy? Maybe Strava is using miles and time? Or some very different treatment of the HR data....

Seattle Forrest

​​​​​​You rode for 4 hours. It gave you 1,400 extra calories for that. Or 350 per hour. I don't know what your BMR is, or even what a reasonable number is, I feel like it's probably less than that, but I think that's what Strava is telling you. You put 1,250 kJ into the bike, and during that time you burned 2,650 calories total. You spent half of them on making the bike go, and the other half plus on running your brain and liver and everything else.

I can't think of anything else that makes sense for them to be doing.

Seattle Forrest

I don't know if this is what you're asking but my understanding is if you upload a ride with nothing but raw location data, Strava calculates how much power it took to move your bike and body at that speed over whatever elevation ... and then uses those power numbers like real ones to guess calories.

MinnMan

That could be it.

bblair

Using the above formula, I get 10cal/minute, or 600 per hour. Looking at a ridewithgps ride, 40 miles of rolling hills in 2:45 that is almost exactly what I get. So, yea, that is the formula they use also.

I use the numbers not for absolute burn, but for relative amounts comparing other rides. Also, using the easy-in-my-head metric of 150cal/beer.....well, you get the idea.I like that better than Zwift's pizza slices metric.

Carbonfiberboy 

I've always had Strava Premium. Before I got a PM, I used Strava for calorie information, simply using their estimate of kJ. Later, using the PM on the same route, kJ were 15% less than the old estimate. I'm probably more aero than they allowed for. W/o PM data, Strava calculates kJ from speed and gain using the standard formulas. Strava's pretty good at that as long as there's little wind and your input data is correct. HR calorie "calculations" are worthless IME. My theory is that marketing told the engineers to jimmy the formulas so that it would come out high and the user would be impressed with how wonderful they were and therefore how wonderful the watch or whatever was. The other possibility of course is that some devices like Wahoo add one's BMR to their (worthless) estimated kJ derived from HR, so then you have no idea of your kJ or calories for that matter.

Case: 154 mile rides uploaded to Strava, sequential years, identical average speed, with PM 4189 kJ, without PM 4866 kJ. I ingested about 3000 calories. Using HR, TP calculated 2534 and 2550 calories respectively.

My TP HR calories are always lower than my Strava kJ, (unlike Strava where their calculated kJ were more than their measured kJ), but if I moved my HR zones down, I'd have crazy amounts of HR Z4. OTOH, maybe I really do ride that hard. I don't know.

WhyFi

I've never seen Strava include base metabolic rate in their calorie estimate.

Like I mentioned before, I think that the power/KJ and calories are being estimated independently. Power/KJ based on course/elevation, speed, rider/bike weight, and assumptions about rolling resistance, position, etc, based on the bike type indicated in the bike profile. Calories based on duration, HR, and at least gender, weight, etc.

To test this, I did as I'd suggested earlier - I went back and looked at a ride without power and changed among three of my bike type profiles: super heavy MTB (I used this for when I'm towing my kid around), gravel bike (which is what the ride actually took place upon) and lightweight road bike.

Super heavy MTB:



Gravel bike:



Road bike:



You can see that the calories don't change despite the significant changes in estimated power/KJ.

MinnMan

Wow - those are huge differences. I'll experiment and see if I see similar. Good work.

WhyFi

Just remember that you need to refresh the activity page after you make changes to the bike profile.

genejockey 

That Calorie count seems awful high for the distance, average speed, and climbing. For comparison, here's a similar length ride with a bit more climbing, at a much higher speed:


Note how close the estimated kj are to estimated Cal.

WhyFi

The calorie estimate is actually low. This was in the winter, in full winter gear, though cold winter air, on carbide-studded 38mm tires that roll like bricks. I have tens of thousands of miles outdoors with a power meter and ~600cal/hour is working at a mild pace for me. An HR of 136bpm is typically going to put me in the 230w neighborhood.

genejockey 

Okay, but how much of that does Strava know, and what, of what Strava knows, does it incorporate into its Calorie calculations? I'd be interested to know whether Calories track kj better when you are using a PM.

WhyFi

That was the whole point of that post - to see if we could eliminate the possibility that their power/kj estimate is influencing their calorie estimate. Yes, calories track much more in line with reality when riding with a power meter.