Fitbit Workout Calories Inaccurate?

[Breeder]

I wear my Fitbit Charge HR and a Polar FT80 when I workout in the gym. There was a discrepancy in the calorific expenditure between the two devices:

Fitbit: 1518cals

Polar: 518cals

There is 1000cals difference. The Polar figure is more accurate from what I can deduce. Anyone comment on this. I'm following the TDEE method for weigh loss.

[I-train-hard]

Daffyduck is correct. All is just an estimate.

[Heybales · Accepted solution]

---

@Breeder wrote:

I wear my Fitbit Charge HR and a Polar FT80 when I workout in the gym. There was a discrepancy in the calorific expenditure between the two devices:

Fitbit: 1518cals

Polar: 518cals

There is 1000cals difference. The Polar figure is more accurate from what I can deduce. Anyone comment on this. I'm following the TDEE method for weigh loss.

---

Since calories is merely calculated from other stats and HR that is seen - what are the actual HR figures, like avgHR and maxHR seen during the workout for both?

Many find the Fitbit doesn't see HR correctly at higher levels.

Now of course your caloire burn would seem to be the other direction from that issue.

In which case - why would you deduce the Polar is more accurate, merely because it's lower?

Since you have the nicer Polar that has a VO2max self-test and figure you can change - have you run the self-test lately after a rest day to confirm figures being used?

Because many have reported the Fitbit matches right up with the nicer Polar's.

Because Fitbit is doing exactly what Polar is doing.

Using a resting HR figure, but not from single test, but from life (and also tad higher which could skew results).

Watching weekly workout time, rather than Polar asking you for workout activity level.

Using BMI figures for estimating VO2 level with above and calcluted HRmax figure.

[DavidSchneier]

Hopefully this isn't shocking to any of us, but definitely not to me:

The point is, you’re not burning as much as you think you are.

I've been subtracting at least 600 calories from my FitBit estimates since early February, which is the only reason I've been able to adjust my calorie consumption to lose weight. If you are counting calories and believe you've created a realistic deficit without the accompanying, expected results, here's your smoking gun.

[Mukluk4]

I would agree.  Each of these trackers assumes a "normal" and generalized person.  In reality, we are all different.  Some people can eat bacon and drink whiskey every day and live to be 100, others die of a heart attack in their 50's doing the same thing.

The point is that you need to get out there and move your **ahem** and eat mindfully.

[JenniferinFL]

I know this is already solved, I would just like to point out something that tends to get forgotten. 

Fitbit lumps your BMR calories in with your exercise. I'm not entirely sure how polar does it, but, I wonder if that isn't the difference. Fitbit spreads out your bmr in 5 minute increments across the whole 24 hour period. 

For me, Fitbit gives me 6 calories every 5 minutes as my BMR. If I go for a half hour walk/job and burn 200 calories, 36 of those calories are BMR calories and 164 are actual activity calories. 

Obviously though, 1000 calories is a lot, for me that's 13.8 hours worth of BMR. You probably are not spending hours in the gym. Though, I dunno, if your Polar is showing a 518 calorie burn you probably are. That seems like a really drastic difference. Thought, it would all be explained away if you just got up in the morning and went straight to the gym and took your overall calorie burn to be your gym burn, because it would include all of your bmr calories since midnight. 

I guess, where are you getting your 1518 calories from? Are you actually going to the activity tab and scrolling down to where it will designate it as an activity and getting it from there or are you just looking at total burn?

[Mukluk4]

So I track my workouts pretty religiously.  Prior to April, I had an Alta (non-HR).  I've since upgraded to a Charge 2.

I went back to compare my morning basketball activities (same court, same people, similar duration) to get a comparison.

Here is from the Alta;

Mar 23, 5:53AM

Basketball

6,655

N/A

1:09:07

679 cals

 (10 cal/minute burn)

Here is from the Charge 2

Today, 5:52AM

Basketball

7,179

N/A

1:07:51

1,146 cals

(17 Cal/minute burn)

Basketball while is part a step based activity, it does not really account for the jumping, sudden start/stops, variation in speed that can occur in seconds, or a defensive stance.  

Based on how the step based trackers calculate calorie burn, I've always felt that it is low.  There is NO way that I can burn 13 cals/minute on a dog walk (not sweaty), 15 cals/minute in a jog (somewhat sweaty), but only 10/cals a minute playing full court basketball (took a bath in sweat).

Skip ahead to the HR.

I took the same route for a jog and it came back with 16/cals a minute, 14/cals a minute for a walk and 17 cals/minute for basketball

So does it over estimate? Maybe but if set up right, I don't think it's a significant amount.

[SunsetRunner]

There is little point in using the calories measured by most if not all of the fitness trackers. A recent study showed an error rate on average of 20% or greater which is significant enough in my opinion to make the calorie tracking completely useless.

Quote below from a BBC news article

"Not one of the devices had an error rate below 20% - and some, such as the PulseOn,were much more inaccurate, the US researchteam found."

[GershonSurge]

@SunsetRunner,

The Fitbits have consistent errors. My Surge is 10-15% over. My Zip is 10-15% under. My Charge 2 is about right. Many of us have learned how to make this adjustment based on calories consumed and weight lost. 

Articles by uninformed journalists are useless. I'd like to see their scientific tests over a month instead of for one exercise.

[Bill_k]

To be honest i don't trust any 'studies' these days. Too many studies with agendas . In the end,yes,  whatever works,works.

Like Mulder says.......

[GershonSurge]

@SunsetRunner,

I collected this data last year when I was losing weight. On the Fitbit Surge, I found it took a 3,927 calorie deficit to lose a pound. The "standard" is 3,500. I just maintained a higher deficit to account for the Surge overestimating calories burned. The R-squared of .9887 means 98.87% of the weight loss was related to the calorie deficit. It doesn't get much better than this. (If an instrument is consistently wrong and can be calibrated, it is considered accurate.)

Compare this to 60 volunteers at Stanford running, walking, and cycling once and comparing the calories measured to the calories burned and then saying the trackers don't work.  The fact that this useless study seems to have been released to many sources at once suggests all of Dr. Ashley's work is equally shoddy.  I'm surprised Stanford didn't bury this poorly constructed study. 

[A_Lurker]

I'm with @GershonSurge  on this.  Using detailed food tracking and using actual loss I can say that my Charge HR is about 10% high.  That's given me enough information to stop the precise tracking and still lose weight (slower, but that was the point).  I do think it's going to be slightly different for every person as none of us are the 'average' BMR.

For most people weight loss isn't easy.  Or perhaps they are the people who only talk about it.  The data is close enough that I do believe most people can use the information - how they use it will be different for each person.  This of course presumes that the person has a metabolism in the (relatively) normal range.

I've also noticed that people (and I'm using people at work, this is not board specific) who don't do well with weight loss are the people who lack the commitment to make it work.  Since I've lost a noticeable amount of weight I've had people ask for advice.

1. track everything (on paper) that you eat for at least 2 weeks (don't change your habits)

2. see if there are obvious quick fixes (think liquid calories)

3. If not, sit down and work out the calories

4. Make a plan

With this I suggest more home cooked foods / prep work so they don't rely on packaged and take out food.  Some take the suggestions to heart, others respond:

too hard

too much time

don't have time

foods too boring (this one I don't get - I eat a variety of items)

And I still see some of them in the cafeteria buying 2-serving pops out of the machine probably twice a day. (let's say 250 cals each).  I mean really, they could likely drop the pop and lose a pound a week easily.  I wish I'd had a single bad habit like this when I started.  Unlike myself who can sometimes be tied to the desk all day, most have active jobs where I bet they are way over 10K steps before getting home.

I said it in another thread - you don't need motivation, you need commitment.  I also believe the more successful people are those who treat it as a lifestyle change, not a diet.

[JenniferinFL]

---

@GershonSurge wrote:

@SunsetRunner,

I collected this data last year when I was losing weight. On the Fitbit Surge, I found it took a 3,927 calorie deficit to lose a pound. The "standard" is 3,500. I just maintained a higher deficit to account for the Surge overestimating calories burned. The R-squared of .9887 means 98.87% of the weight loss was related to the calorie deficit. It doesn't get much better than this. (If an instrument is consistently wrong and can be calibrated, it is considered accurate.)

Compare this to 60 volunteers at Stanford running, walking, and cycling once and comparing the calories measured to the calories burned and then saying the trackers don't work.  The fact that this useless study seems to have been released to many sources at once suggests all of Dr. Ashley's work is equally shoddy.  I'm surprised Stanford didn't bury this poorly constructed study. 

---

I was just talking about that study with someone. No activity tracker can be accurate for calories burned for everyone, but, some of them can be precise. Basically, they were comparing the activity trackers to medical grade expired gas measurement using indirect calorimetry and then saying "Ha, these are not accurate." Of course they are not. They cannot be, the only way to have an accurate calorie burn calculation is if you feel like spending your whole day breathing into a mask so that the expired gases can be measured by real hospital-grade medical equipment.  But does it really matter that they are not accurate for everyone? No, not at all. 

A device can be precise without being accurate. For instance, when you weigh in on a bathroom scale, there is a good chance that your weight is not accurate. A scale that is accurate is usually quite expensive. A scale that is legally allowed to be used for trade is going to be a lot more expensive than your bathroom scale, but, that's where you would find an accurate scale. Your bathroom scale is useful to you because it is relatively precise. It might be a couple lbs off from your real weight, but, so long as it consistently shows a higher weight when you gain weight and a lower weight when you lose than it is still a useful tool. If your scale is consistently 2 lbs off from the scale at your dr's office, you can still tell that you lost 10 lbs by using it consistently. 

What was interesting was this study showed that the fitbit was always within 5% on the heart rate. That's pretty precise for a wrist worn device. The fitbit is really just a collector of data, heart rate and steps. After that, a calculation is run using accepted standards to determine how many calories were burned. Even if you had two candidates who were identical in stats, they would burn different amounts of calories doing the exact same activity. That's why it can take people some time to find out their actual TDEE. It's not a perfect science. But, there is no better alternative because nobody can wear an indirect calorimetry machine all day, everyday. 

Besides, just like your bathroom scale, a Fitbit can be precise even if it isn't accurate. Mine is definitely precise. It's about 3% off now from my actual TDEE. For me, my Fitbit Blaze is precise and very close to accurate. I think that so long as you aren't doing crazy stuff with it, like tying it on your ankle so it counts your recumbent bike as steps you can have a device that is precise and works with a bit of math. Some people will be lucky enough to also have the device be accurate if they burn the same as the average. 

I just don't understand the point of the study. It was the biggest "No, duh!" study ever. 🙂

[kskipper2]

Fitbit 2 shows probably double calories I burn in a day. Minutes shows at least double amount that I have used. I have reset and may work for a day and I have removed fitbit from account and added it to account again. Again it worked for a few days then went crazy again.

please help

[Heybales]

---

@kskipper2wrote:

Fitbit 2 shows probably double calories I burn in a day. Minutes shows at least double amount that I have used. I have reset and may work for a day and I have removed fitbit from account and added it to account again. Again it worked for a few days then went crazy again.

please help

---

So is that estimate of almost double based on a comparison to other similar activity/workout days that read much lower?

Or in comparison to what you think is a right number?

I only ask because some people have no experience on calories, burning or eating, beyond the 1200 calorie diets they see in articles.

So their own device saying they burn 2400 in a day seems impossible to them - when it's not at all.

Beyond that - there was a bug in the past where despite entering your weight in pounds and displaying it in lbs, Fitbit internally saved it as kg, instead of converting it to kg, for all the formulas.

And that would lead to almost 2 x the calorie burn.

You could actually look on your daily calorie burn graph during sleeping, which is BMR level burn, and do the math and prove exactly what happened.

For that bug, changing to metric, and then back again to lbs, re-entering the weight, syncing device - was enough. I don't recall people saying it came back.

Seemed like a one incident deal, effected a bunch of people, but once corrected didn't happen again.

I guess something really buggy could be up, and it keeps resetting the saved setting to metric.

[passingby]

I've managed to lose about 16lbs in a little over 7 weeks with my Charge 2 HR, but only by significantly reducing the FB proposed food calorie budget estimate because FB way over-estimates the calories burned during activity above BMR.

The basic data collected by the device (steps, HR) is pretty accurate compared with other devices (HR as compared to treadmill cardiac chest strap). The only exercise I do is walking (at about 3mph/20min mile). However, using data export, FB says I should have lost 24lbs over this period of time ((total calories burned – total calories in)/3500). That means either “calories out” is being over-estimated, or “calories in” is being underestimated.

I know my "calorie in" is as accurate as I can get since it's based off of calories listed on food packaging and I ate every bite in a package and not a morsel more. That means the FB "calorie out" estimate is the part that's off.

There are two basic parts to "calorie out" to look at: BMR and burn due to activity above BMR. The FB BMR calculation seems to be in pretty good agreement with other calculators. If you leave your FB sitting on a counter unused for a couple of days, you can see what it's estimating for BMR (no activity) in “calories out” for those days. My computed BMR (Mifflin-St Jeor equation for men: BMR = (10 x weight in kg) + (6.25 x height in cm) – (4.92 x age) + 5) differs from my FB reported BMR by only about 5 calories/day. That's negligible.

So the error, as is consistent with what others have noted here and on other threads in this forum, must be in how FB estimates calories for activity above BMR. I can calculate what my actual calories above BMR for a day was from: (weight change)*3500 = calories in – (BMR + active calories). This isn’t very accurate for a given day where weight change depends more on what you ate or evacuated, but over a period of 7 weeks, the noise in the data averages out and the data becomes useful. For my data I find that my actual daily average active calorie burn was about 416 calories. FB, however shows my active calories (calories burned – BMR) as being about 1140, or 2.74 times larger than it actually was, or (1140-416) = 724 calories/day that I really hadn't burned off.

To get my approximately 2lb/week loss, I told FB I wanted to lose 2 lbs a week, which adds about a 1000 calorie/day deficit. I then aimed to end the day ~700 calories under the FB food calorie “budget”. This ensured that I was actually hitting the 1k/day deficit and achieving the desired loss rate.

If I had used the actual FB calorie goal as a target, I’d have only been burning an extra 276 calories/day and seeing a 0.5lb/wk drop. If I had set my goal to maintain weight, I would have been taking in an extra 1300 cal/day, and gained 2-3 pounds a week.

As noted elsewhere, the FB active calorie estimate is way off, crediting you for 2-3 times the number of calories you should get for walking at 3mph. giving you credit for far more calories than your activity actually burns.  This is after removing BMR from the equation.  If FB “learns” as it goes along to make better estimates, it doesn't seemed to have learned much in 7 wks of daily data collection.

This problem has existed for years, and FB does not seem to have any intentions to correct it.  A rule of thumb/sanity check is that if your “calories in” match your BMR (provided by FB per above or a web calculator), then the only weight loss you’re going to have is from active calories. Tweak "calories in" up or down from there to get your weight to do what you want it to do.  If you take the FB target calorie estimates at face value, which include their estimated active calories on top of BMR, you're going to be sorely disappointed.

[RLSteward]

Mine seems to be an extreme issue. 

It has recently been coming up with calories in the 100,000s for the same activities over the last week by about 71 times. Nonsense. 

Recent Feb stats:

9 Feb  3124

10  3862

11  3501

12. 3850

13. 248,936

14  248,936

15. 3454

16. 2749

17  248,936

18. 248,936

What does this mean?

Thx,

Robyn

[Dominique]

Is the issue only with calories, or also with steps, active minutes etc.? Have you try to restart your tracker?

[RLSteward]

Yes, seems to be. Although when i cycle my runkeeper stats and Fitbit Alta 2 stats  vary a small amount.

No how do I do that?

[Dominique]

How do I restart my Fitbit device?