Fitbit's calorie counting algorithm

[DoctorCoral]

First off, this is not a criticism of my Fitbit Versa 2, but an honest question.

As a scientist, I have a type A personality, and, even before I had a Fitbit, I logged every calorie I consumed. Until my Fitbit, though, I had no idea how many calories I BURNED, but I suspected that I was usually at a net deficit (more calories burned than consumed). As it turns out, not only is that the case, but it's not even close: I burn 1,200-1,300 more calories/day than I consume. I am 183 cm and 105 kg with a BMI of 25-27 (i.e., fat). My target weight based on my frame size is ~75-80 kg. I accumulate my recommended 150 minutes of "active minutes" for the week by noon on Monday (i.e,. I exercise ridiculously intensively via biking, swimming, or running). Now I know it's not all about calories, and it matters WHAT you eat, but come on: if I drank 2,000 calories of lard/day and burned 3,000 calories.....wouldn't I lose weight just in terms of shear physics (i.e., laws of thermodynamics)?! How can you gain weight by burning 1,300 calories more than you consume a day? It seems to be violating a law of physics unless I am converting water into energy, which, as a biologist, I do not think is possible. 

What I'm wondering is this: Fitbit is calculating calorie loss based on my age, gender (male), size, and heart rate. Is it possible that it's just completely wrong? For instance, maybe I ran 7 miles, and it THOUGHT I burned 1,200 calories based on my heart rate, but, because I exercise so much and so often, I actually only burned 500. Then, it APPEARS that I burned 3,800 calories/day (my average since July 2020) but in reality, I burned maybe 2,500 (around the amount I consume a day), in which case that COULD explain my lack of weight loss. Otherwise, if I burned 1,200 calories/day more than I ate since July 8th 2020, I should have lost 120,000 calories=~40 pounds (18 kg), when in reality I gained a few pounds. Does anyone know why the Fitbit calorie estimate could be THAT far off? And is there a way to tailor the algorithm to your (potentially sluggish) metabolism? Do I simply have a metabolic disorder? If I burned 120,000 net calories (over amount consumed) in four months yet gained weight, I feel like I should participate in some sort of study because my body is clearly doing something that is far enough from normal that may make me of interest to human physiologists! More likely, though, the Fitbit calorie counter is just off by a factor of nearly 1.5-fold (150%). 

[Mughal]

the calorie count for green lentils is twice what is shown on calorie counts  on internet.  Also one of the choices for chickpeas has huge amounts listed.   There is no fresh spinach shown only frozen, cooked and then measured.  Amounts for a lot of vegetables are for chopped in cup rather than whole in a cup.

[elstar]

I can't help you with the Fitbit algorithm (I really wish they'd design it to use your weight data to make the weight loss calculator a little more accurate!) but I can tell you that I found what worked for me using the calorie recommendations of SparkPeople. I tend to have pretty good results adding my purposeful activity calories burned to my recommended calories for a totally sedentary day number. Instead of the 2200 calories recommended by Fitbit, I'm closer to 1800-1900 and it's working great!

[Heybales]

So besides the 1st week of weight loss going into a diet where everyone loses extra water weight and therefore no calories - you have been losing 2.5 lbs weekly?

That would be the case if burn is about 1250 more than consume.

When you say always at a net deficit, was that because logging food was part of a diet, or curiosity?

If not losing fat weight then you really weren't.

As to what Fitbit uses for calorie burn. And it's intended decent estimate for average uses cases - and way off if you are not.

Daily life is your estimated calculated BMR, plus any steps make a distance, and distance and mass is very accurate calorie burn.

Therefore underestimated in your day is any non-step time awake (because you do burn more than BMR), and standing time (much more than BMR).

But that balances out bogus steps with tiny distance and calorie burn.

Where that balance is for you could be decent, or perhaps a very long standing work day moving arms but no steps and badly underestimated.

Exercise is by HR calculations.

HR-based calorie formula is only valid for steady-state aerobic exercise (same HR for 4-6 min), and even that range is inflated at bottom (above daily activity) and at top (below anaerobic).

Anything anaerobic or with HR all over the place is inflated calorie burn in that formula.

So if a lot of weekly workouts and time are nothing but calisthenics (true term for fad HIIT slapped on all kinds of non-HIIT stuff), or lifting - then badly inflated calorie burn.

If a lot of walking time which would be much more accurate by distance & mass bumps up into HR-based calorie burn - inflated.

And even if not, if a lot of steps and stride length is wrong, that can cause it to be off.

And even HR-based calorie burn used for the right kind of workouts - the formula's are limited as they are assuming some key values - VO2max and HRmax.

HRmax is calculated at 220-age and may be no where near correct.

Imagine the calorie difference if doing workouts avg HR around 160 bpm - if the Fitbit thinks your HRmax is 168 (95% of HRmax anaerobic range if true), but it's actually 190 (84% right in aerobic range).

That's my case.

VO2max is calculated on restingHR, HRmax, BMI, gender, and exercise stats.

restingHR can be known, but Fitbit's value for it (unless they have a true 1st thing in morning hidden value) includes your daytime awake resting also.

HRmax problematic as above.

BMI, gender and stats is along lines of if BMI is bad then fitness level must be bad.

But we probably know plenty of healthy weight people terribly unfit, and many "overweight" by BMI that are very fit - formula like that weights that decently.

The other side of this equation is of course the food going in.

You said careful logging - that means different things to different people.

As far as nutritional labels are concerned and what happens in body - calories is per gram, weight, not by volume, cups, spoons, "about 3 servings per package".

Do you weigh everything you eat but liquids?

Do you confirm the database entry matches the label in your hand for those kinds of foods?

Do you use a USDA entry for fresh foods?

How often do you eat take out/delivery that is going to be very inaccurate?

From your mass of workouts done - for sure your calorie burn is overstated.

If you have a lot of steps outside the workouts - could be adding to problem if stride length has never been confirmed.

Eating side could be adding to the problem.

Also - what started as a reasonable deficit for the fat available to lose, could be rather extreme at some point and body was not happy with deficit being thrown on it - and it has adapted to slow you down in subtle but meaningful ways to burn less than you were.

Usually with that kind of stress on body is increased cortisol - can gain upwards of 20 lbs water weight that way - that could hide how many weeks of continuing fat loss?

And obviously if body is stressed that way - not great state to be in.

That's usually when the body will get what it needs somehow, like unable to fight off sickness - nothing like total rest for a week to slow you down!

Or injury - nothing like 6-8 weeks greatly reduced activity to catch up!

Much better to purposely reduce the deficit with less to lose, than have the body force it on you.

How much weight until healthy weight you have left?

[DoctorCoral]

I basically double the calories in everything I eat to make sure I'm not undercounting, so I think what is happened is one of the final points you made: 

"Also - what started as a reasonable deficit for the fat available to lose, could be rather extreme at some point and body was not happy with deficit being thrown on it - and it has adapted to slow you down in subtle but meaningful ways to burn less than you were."

Basically, my body is now used to a 1,300 calorie deficit/day. So if I have a deficit of 500, I gain weight. It still doesn't make sense to me in terms of thermodynamics (how can you gain weight by net-burning calories?), but if the calorie counting algorithm is way off for me, then maybe my purported 1,300 calorie deficit is more like 650, and if the calorie data on the packages (what I use for my estimates) are way off, then maybe I think I'm eating 2,500 calories/day when it's more like 3,250, i.e., no caloric deficit.

For some activities, I feel like it actually underestimated my calorie consumption: 30 minutes of running at 230 lbs at a mean heart rate of 140-150 bpm=only 400 calories! My VO2max is about 48-50, so not terrible or great. Maybe I can find a different means of calculating calorie burn and see if there is bias in the Fitbit algorithm for someone of my frame/size and metabolism. Thanks for the suggestions. 

[Heybales]

And be aware that fat is not fast lost or gained.

If you eat 500 calories over maintenance daily in reality, it would take 1 week to slowly gain 1 lb.

If you quickly gain 1 lb in couple days - it's water weight from finally restoring some glycogen stores that store with water.

Not fat.

Same as people that are in a deficit, you could be losing 1 lb weekly implying a 500 cal deficit, eat 250 more daily, or 500 more daily, for like 3 days and gain a couple lbs. Obviously water weight, since not even eating above maintenance.

It's not that your body is used to 1300 cal deficit - you don't actually have it because for it to stop the foolishness you are now probably burning less through various means.

So indeed, unless water weight is involved from stress, results trump logged numbers.

And how you got to this point dictates likely water weight or not.

If someone had a reasonable 1 lb weekly loss rate on paper and it showed up after 4 weeks as 3.8 lbs or 4.3 lbs - that's showing their logging is off somewhere.

If someone attempted 4 lb weekly rate and it was never reached and over weeks the loss stopped, probably some logging - and water weight from stress.

For the former, adjust the numbers to compensate.

For the latter, slowly get out of the diet and reset for a week. Tackle it again with reasonable rate.

For compare of calorie burn, gross is option that should match Fitbit.

https://exrx.net/Calculators/WalkRunMETs

Yes, should be more than 400 cal probably.