Apple Watch calorie estimates…

[Deguello]

Is there any way short of wildly misrepresenting my weight to get the watch to give calorie estimates closer to reality? It’s fairly precise, but its accuracy is quite poor. It overestimates my TDEE by about 700 calories a day.

Yes, I can take that error into account. Yes, the trend might be more useful than the absolute number, but my calorie burn is very consistent day-to-day so there isn’t much of a change except gradual increases or decreases as I gain or lose weight.

Most likely, it’s the active calories that are driving the number so far from reality. I’d be surprised if my resting calories were wildly out of sync with the default calculation.

 

[Howard2k]

How do you know it's overestimated your TDEE?

 

[Deguello]

Math.

I maintain at around 3400 calories and the watch says I burn around 4100.

 

[Howard2k]

So you're saying that if you eat 3,400 calories of food, your weight stays the same, and therefore the watch is wrong?

 

[Deguello]

Yes.

Over time, 3400 calories a day keeps my weight within a very small range. At 4100 I’d be gaining more than a pound a week.

 

[Howard2k]

If it helps, both your 3,400 weight maintenance theory, and Apple's absolute calorie metrics will be both wrong to some extent.

Your idea is wrong because the way your body burns calories (and your metabolism) varies based on even simple things like how you slept, the outside temperature, differences in exercise intensity etc. Even if you ate the exact same portions of the exact same foods at the exact same time every day (and you don't) this would still not be super precise.

And the watch's active calorie estimates are also not going to be super precise because they're based on big data analysis. Though the watch is almost certainly closer than your formula. I say almost, but really I'm quite certain.

I would accept that both are wrong, but the Watch is going to be closer to reality.

Having said that, no, aside from playing with health metrics (age, height, weight etc) I don't know of a way to tune the algorithms on the watch.

 

[Deguello]

Yes, my nine years of logging food is probably off to some extent. Not 700 calories a day, though.

My idea isn’t wrong. it doesn’t have to be super precise. It had to be relatively accurate over time. And it is. This is shown by nine years of scale data.

The watch absolutely isn’t closer to reality than the combination of my logging and my scale. Not even close.

 

[Howard2k]

I'm not knocking your idea at all, I think it's a great idea. How did you factor in aging and how old are you out of interest? Again, back to the original question, I have nothing further to add, just curious. It's always cool to hear how people use health metrics. For a while I was measuring my weight and body fat (estimated on a scale) and I'd import that into a spreadsheet each day and put it on my fridge. It's always cool to find new ways to use data.

 

[Deguello]

My age doesn’t matter. It might matter to a formula. It doesn’t matter to my scale or my logging.

Right now, at my very consistent level of burn, I maintain on 3400. When the weather turns cold again and I don’t feel like going outside as much, I’ll maintain on less. Probably around 2800-2900. And the watch will overestimate that by around 600.

 

[G46&Fbnth5]

Reducing your weight in the app should decrease the burnt calories detected by the Watch.

Anyway devices like Apple Watch are not reliable and accurate enough to track calories, if you have a strict schedule.

 

[Deguello]

Yeah, I know reducing my weight will reduce my estimated calories, and I might try that just to see how much of a reduction it would take.

The watch is very consistent — precise — but either its baseline or algorithm, or both, for turning my activities into calories. My daily activity doesn’t vary much and the watch consistently puts the total within a range 100 calories or so at this activity level. It’s just that the total is off by about 500 at a casual level of activity and the difference widens the more I exercise.

 

[BlackBun]

I’ve always thought the Watch generous in determining calorie expenditure but like the VO2 max, you’ll need to book a lab slot and hook yourself up to a machine.

Just looked at this: I’ll go with my AW as this website was wayyyeeeeed too generous.

Calories Needed for a 63 year old, 65 kg Male

A 63 year old male who is 65kg needs an estimated 2,314 calories per day

tdeecalculator.net

All bollox any way.

 

[Deguello]

I know, based on real world results, about what my TDEE is at this activity level. I was just hoping there was some way to “calibrate” Apple’s estimates but I’ve made it this far without such a thing and I’ll be fine with its overestimate.

 

[akash.nu]

I’ll counter that argument with my own observations. I’ve been using an  WATCH since the series 4 onwards and it seems to have been showing results that’s within tolerable error of margin for me. At a max 100 calories I think.

 

[Deguello]

Of course the algorithm is going to be somewhat accurate for some people. That’s zero help for those for whom it isn’t.

 

[Indianwin2001]

Of course the algorithm is going to be somewhat accurate for some people. That’s zero help for those for whom it isn’t.

I have both the AWU ands a Garmin. They both calculate about the same number of active calories during workouts, with the Garmin giving a few more

 

[Deguello]

Yes, they probably use similar algorithms. That doesn’t mean they fit everyone.

 

[randomando]

Could be a combination of different things. Have you measured resting or active calories through other devices? (Not just through how much you are eating).

 

[Deguello]

What combination of different things do you think it can be? If I know what my intake is and its effect on my weight, I have my answer.

If I eat, over time, 3400 calories a day and, over time, my weight stays stable, the answer to how many calories I burn a day is 3400, regardless of what the watch or any other device says.

 

[Brandon42]

Slightly different situation than yours but I have wondered about the accuracy of exercise calorie counting in special situations. For example:
18 minute mile walk: ~75 active calories
18 minute mile walk with short bursts of brisk walking: ~130 active calories.

Yes the short bursts bump up my heart rate but the total distance and total time hasn’t changed but is approaching a 100% increase in active calories.

Just seems too good to be true.

 

[Brandon42]

What combination of different things do you think it can be? If I know what my intake is and its effect on my weight, I have my answer.

If I eat, over time, 3400 calories a day and, over time, my weight stays stable, the answer to how many calories I burn a day is 3400, regardless of what the watch or any other device says.

It is a bit more complex than that but I would agree that it should be close on average. Metabolizing those calories depends on what the source of calories is. Also, as I understand it, there is research showing that you burn about 15 calories per serving of water (think of the energy you burn to bring the water up to body temp for example). Lastly, if you add muscle from exercise that is going to throw off your calculations.

Personally, I would (and I do) keep a decent estimate of calorie consumption and keep an eye on weight and active calories. If your weight moves away from where you want, adjust your active calories or calorie consumption. The exact numbers don’t matter, the changes do matter.

 

[Deguello]

It is a bit more complex than that but I would agree that it should be close on average. Metabolizing those calories depends on what the source of calories is. Also, as I understand it, there is research showing that you burn about 15 calories per serving of water (think of the energy you burn to bring the water up to body temp for example). Lastly, if you add muscle from exercise that is going to throw off your calculations.

Personally, I would (and I do) keep a decent estimate of calorie consumption and keep an eye on weight and active calories. If your weight moves away from where you want, adjust your active calories or calorie consumption. The exact numbers don’t matter, the changes do matter.

Yes, protein has higher thermic energy than carbs and fat but, over time -- and I'm talking about years, here -- that kind of relatively small difference sorts itself out. And 3500 calories per pound is an estimate. And doing more exercise might reduce NEAT for some people and cause more NEAT for others.

Regardless, it's a more accurate approach than an algorithm that estimates based on on various physical factors other than intake and weight changes.

I keep a very detailed record of calorie consumption and its effect on my weight. I have a very solid handle on how that works. I was just wondering if there was a way to get the watch to reflect what I actually burn rather than running it through Excel and trying to massage it. I can do that -- I am doing that -- and the more activity I have the further off the watch is. That's why I think it's about the active calories not the resting ones.

 

[Deguello]

Slightly different situation than yours but I have wondered about the accuracy of exercise calorie counting in special situations. For example:
18 minute mile walk: ~75 active calories
18 minute mile walk with short bursts of brisk walking: ~130 active calories.

Yes the short bursts bump up my heart rate but the total distance and total time hasn’t changed but is approaching a 100% increase in active calories.

Just seems too good to be true.

I don't see that much variance, but I don't doubt it. My most consistent exercise to compare is a 4.17 mile walk I do in 56-57 minutes and that I've done every day for 1400 days or so in a row. The watch has logged a range of 620-720 calories or so for the same route. I can buy that because some days really do take more effort to maintain the same pace. The days after leg day or toward the end of a weight cut take a lot more effort. Sometimes doing it fasted is harder.

I'm pretty happy with the precision of the watch. It's surprisingly consistent and my typical days fall within a very narrow range of estimated burn. It's just not accurate. My 60 rolling TDEE is 3424. The watch says 4181. Adjusted with a formula I have in Excel based on all the data I have since I bought a 4 watch, it's at 3379 over the same 60 days. That's less than half a pound of difference per month (a couple of glasses of water worth of retention or a little extra salt, even) and both the 3424 and 3379 are very usable.

 

[Brandon42]

One random thought: if you take beta blockers or calcium channel blockers, that has an impact on heart rate. There is a setting to adjust for that in the Health settings. I don’t know if it has an impact specifically on the active calories algorithm though.

 

[TechnoMonk]

Pretty much all calorie estimators are useless/unreliable. I didn’t realize how wildly inaccurate these calorie estimators were until, I was doing weekly BMR/RER testing with breathing in to a mask/pipe while working out on tread mill or bike. What you eat, long term consistent diet also impacts calorie expenditure, not to mention hormonal variations. I had a guy on my team, who was burning predominately fat at 85%-90% HRM. He was mostly Keto except few days of carb cycling, his BMR was slightly higher for his age/weight and HR.