Serious flaw with My Fitness Pal

13

Replies

  • callsitlikeiseeit
    callsitlikeiseeit Posts: 8,633 Member
    your points are nothing new to most of us veterans.

    which is why we nearly always tell people to eat back half their exercise calories, and adjust up or down, depending on their weight loss over a few weeks time (to mitigate any normal weight fluctuations or water retention due to new exercise regime/TOM/ etc )
  • sijomial
    sijomial Posts: 19,507 Member
    sijomial wrote: »
    sijomial wrote: »
    Some of the exercise estimates used by MyFitnessPal source the Compoendium of Physical Activities.
    And that gives estimates in METS which as the OP correctly points out is inflated by 1 MET - a gross calorie estimate not a net calorie estimate.

    For short duration exercise it's a trivial difference.
    For long duration exercise it starts to mount up - it's a particularly large percentage deviation for long duration and low rate of burn exercises such as walking.

    Yes the cycling speed categories are badly inflated for any sort of decent road bike - but I don't know any serious cyclists who would use such a dreadful way to estimate calories when much better ones are commonly used!
    There's a world of difference between a mountain bike being riden in tough terrain (or over mountains!) and a fast road bike on the flat which speed ranges don't take into account.

    BTW - please can people stop insisting the exercise estimates were accurate when they mean their weight loss results were on track. That verifies your calorie balance but it certainly doesn't validate all the elements involved on both sides of the CICO balance. Unless you have validated your exercise against a known good method (such as a power meter on a bike) you simply know the estimates worked for you - not that they were accurate.

    Did someone say they were accurate? I guess I didn't read any of the responses that way.

    Same goes for the tracking devices. All are standard algorithms that are used as estimates. Not really accurate, just close enough.

    You said "and yet the MFP exercise calories were right on for me."

    If that wasn't a suggestion that you thought they were accurate maybe you should rephrase?

    "I used the exercise estimates here and lost on track" is a very different statement to the exercise calories were right on for me.

    Not interested in a semantics side argument that is irrelevant to the point.

    Maybe you should re-read.

    Back pedalling might be a decent calorie burner for you. Give it a go.
    :wink:
  • cmriverside
    cmriverside Posts: 32,186 Member
    sijomial wrote: »
    sijomial wrote: »
    sijomial wrote: »
    Some of the exercise estimates used by MyFitnessPal source the Compoendium of Physical Activities.
    And that gives estimates in METS which as the OP correctly points out is inflated by 1 MET - a gross calorie estimate not a net calorie estimate.

    For short duration exercise it's a trivial difference.
    For long duration exercise it starts to mount up - it's a particularly large percentage deviation for long duration and low rate of burn exercises such as walking.

    Yes the cycling speed categories are badly inflated for any sort of decent road bike - but I don't know any serious cyclists who would use such a dreadful way to estimate calories when much better ones are commonly used!
    There's a world of difference between a mountain bike being riden in tough terrain (or over mountains!) and a fast road bike on the flat which speed ranges don't take into account.

    BTW - please can people stop insisting the exercise estimates were accurate when they mean their weight loss results were on track. That verifies your calorie balance but it certainly doesn't validate all the elements involved on both sides of the CICO balance. Unless you have validated your exercise against a known good method (such as a power meter on a bike) you simply know the estimates worked for you - not that they were accurate.

    Did someone say they were accurate? I guess I didn't read any of the responses that way.

    Same goes for the tracking devices. All are standard algorithms that are used as estimates. Not really accurate, just close enough.

    You said "and yet the MFP exercise calories were right on for me."

    If that wasn't a suggestion that you thought they were accurate maybe you should rephrase?

    "I used the exercise estimates here and lost on track" is a very different statement to the exercise calories were right on for me.

    Not interested in a semantics side argument that is irrelevant to the point.

    Maybe you should re-read.

    Back pedalling might be a decent calorie burner for you. Give it a go.
    :wink:

    Negative. Even pedaling forward has proven to be too dangerous for me. :lol:
  • cmriverside
    cmriverside Posts: 32,186 Member
    edited September 2019
    sijomial wrote: »
    sijomial wrote: »
    sijomial wrote: »
    sijomial wrote: »
    Some of the exercise estimates used by MyFitnessPal source the Compoendium of Physical Activities.
    And that gives estimates in METS which as the OP correctly points out is inflated by 1 MET - a gross calorie estimate not a net calorie estimate.

    For short duration exercise it's a trivial difference.
    For long duration exercise it starts to mount up - it's a particularly large percentage deviation for long duration and low rate of burn exercises such as walking.

    Yes the cycling speed categories are badly inflated for any sort of decent road bike - but I don't know any serious cyclists who would use such a dreadful way to estimate calories when much better ones are commonly used!
    There's a world of difference between a mountain bike being riden in tough terrain (or over mountains!) and a fast road bike on the flat which speed ranges don't take into account.

    BTW - please can people stop insisting the exercise estimates were accurate when they mean their weight loss results were on track. That verifies your calorie balance but it certainly doesn't validate all the elements involved on both sides of the CICO balance. Unless you have validated your exercise against a known good method (such as a power meter on a bike) you simply know the estimates worked for you - not that they were accurate.

    Did someone say they were accurate? I guess I didn't read any of the responses that way.

    Same goes for the tracking devices. All are standard algorithms that are used as estimates. Not really accurate, just close enough.

    You said "and yet the MFP exercise calories were right on for me."

    If that wasn't a suggestion that you thought they were accurate maybe you should rephrase?

    "I used the exercise estimates here and lost on track" is a very different statement to the exercise calories were right on for me.

    Not interested in a semantics side argument that is irrelevant to the point.

    Maybe you should re-read.

    Back pedalling might be a decent calorie burner for you. Give it a go.
    :wink:

    Negative. Even pedaling forward has proven to be too dangerous for me. :lol:

    Me too.
    No database entry found for "nursing wounds at the side of the road".

    <snip image>

    Dang.

    I live on a property that abuts a popular recreational urban bike trail. At one end of my property is a hill and at the other end is a blind curve which is narrow and which has black ice in the winter that cyclists can't see if coming from east to west.

    I have been the 911 report caller on a few bike accidents needing EMS: the kind where they go down and don't get back up. Well now people have their own phones, but I used to landline call.

    I got rid of my bike after one too many of my own falls.

    Be careful out there.
  • WinoGelato
    WinoGelato Posts: 13,456 Member
    I'm not talking about MFP is that's irrelevant. I'm talking just about the exercise estimates. I put in very sedentary as my base since my job is sedentary. But i exercise a lot. And every calorie burn estimate on this app significantly over estimates calories burned. I've researched every major category I do typically and the app is if significantly...often it appears to estimate gross calories...and sometimes its just wildly high.

    Is your step count less than 3-5K per day? If not, you aren't sedentary, no matter if you have a desk job or not.
  • ritzvin
    ritzvin Posts: 2,843 Member
    ditto. currently sidelined by a broken collarbone. <sob>
    And a cycling buddy that just got back in crashed again yesterday (thankfully, he was fine this time though).
  • sijomial
    sijomial Posts: 19,507 Member
    ritzvin wrote: »
    ditto. currently sidelined by a broken collarbone. <sob>
    And a cycling buddy that just got back in crashed again yesterday (thankfully, he was fine this time though).

    I'm thanking my Mum for making me drink loads of milk when I was young and giving me strong bones (myth alert!?) as I merely displaced my clavicle.
    Shame the milk didn't make my medial knee ligament strong too. :neutral:

    Get well soon.
  • ritzvin
    ritzvin Posts: 2,843 Member
    thanks. I'm hoping for a quick recovery - I miss riding so much and road season will be pretty close to over by the time I'm likely to recover. Per the X-ray, it's lined up fairly well and may not need surgery unless next X-ray shows otherwise.
  • cmriverside
    cmriverside Posts: 32,186 Member
    Oh man. I broke a collarbone in a bike fall too. Brutal.
  • ritzvin
    ritzvin Posts: 2,843 Member
    Oh man. I broke a collarbone in a bike fall too. Brutal.

    We belong to a rather large fold. It's such a common cycling injury (almost just a matter of time when I realized just how many of my fellow riders have as well).
  • jjpptt2
    jjpptt2 Posts: 5,601 Member
    I agree with both forks this conversation has taken.

    This isn't highly exacting process that takes mental Olympics to manage. Good enough can certainly be good enough.

    However...

    Managing all yoru estimates in a way such that the error rates are managed (or canceling each other out) doesn't mean you are being accurate.
  • sijomial
    sijomial Posts: 19,507 Member
    kshama2001 wrote: »
    sijomial wrote: »
    kshama2001 wrote: »
    sijomial wrote: »
    Some of the exercise estimates used by MyFitnessPal source the Compoendium of Physical Activities.
    And that gives estimates in METS which as the OP correctly points out is inflated by 1 MET - a gross calorie estimate not a net calorie estimate.

    For short duration exercise it's a trivial difference.
    For long duration exercise it starts to mount up - it's a particularly large percentage deviation for long duration and low rate of burn exercises such as walking.

    Yes the cycling speed categories are badly inflated for any sort of decent road bike - but I don't know any serious cyclists who would use such a dreadful way to estimate calories when much better ones are commonly used!
    There's a world of difference between a mountain bike being riden in tough terrain (or over mountains!) and a fast road bike on the flat which speed ranges don't take into account.

    BTW - please can people stop insisting the exercise estimates were accurate when they mean their weight loss results were on track. That verifies your calorie balance but it certainly doesn't validate all the elements involved on both sides of the CICO balance. Unless you have validated your exercise against a known good method (such as a power meter on a bike) you simply know the estimates worked for you - not that they were accurate.

    Did someone say they were accurate? I guess I didn't read any of the responses that way.

    Same goes for the tracking devices. All are standard algorithms that are used as estimates. Not really accurate, just close enough.

    I said the particular walking entry I use was accurate for me (Walking, 3.0 mph, mod. pace, walking dog)

    Because I eat a lot of whole foods and get my syntax for those entries from the USDA database, I am very confident in my food logging but was having trouble losing weight until I changed some exercise choices.

    I gave strength training examples above.

    For gardening, I give myself credit in a range from no credit to full credit. If there's a shovel involved, I take full credit. If I'm clipping roses like Annette Bening in "American Beauty", I take no credit. (However, if I am pruning up a storm, that's different.)


    Silly picture removed.

    Same question - how did you validate that the walking estimate was accurate for you?
    Hooked up to a breath analyser in a gym perhaps to measure oxygen uptake?

    How did you validate your BMR was accurate?
    How did you validate your activity multiplier was accurate?
    How did you measure your TEF?

    Again getting the right results doesn't not validate all the elements involved.

    e.g.
    When I started out I used inflated exercise estimates from a HRM and gym machines.
    Adjusted my daily goal and proceeded to lose 1lb a week as predicted - my exercise estimates were still inaccurate, losing exactly on goal didn't change that, I just compensated for the inaccuracy.

    Losing as expected over time is all the validation I need that I am using good entries.

    I'm afraid people will read your post and think weight loss is a big complicated technical process, when our bodies can handle these processes. We just need to burn more than we eat. We don't need to be perfectly, excruciatingly accurate for this to happen.

    I would hope they would realise that reasonable estimates in all aspects and the common sense to make adjustments based on results is what's actually required. Not chasing perfection or absolute accuracy.

    Look how simple my example was. Logged my food completely but not absolutely accurately, used easily available exercise estimates, adjusted my calorie goal based on actual weight lost over time = success.

    For me as a high volume exerciser (like the OP appears to be) investing some time in making my exercise estimates fairly accurate made sense - for a low volume exerciser it's a small part of the puzzle.

  • jenilla1
    jenilla1 Posts: 11,142 Member
    I see people warn about this frequently, but it just hasn't been an issue for me. I've been here maintaining at goal for over 7 years now. When I was losing, it worked fine, too. This hasn't been a problem for me at all. The calories MFP has given me are very close to what I get with my Fitbit. In fact, MFP often underestimates for me, personally. It must not be that far off from reality if I've managed to stay within a few pounds of goal for several years now. It's doing something right...

    P.S. I'm set to sedentary, and I add in all my exercise calories.
  • cmriverside
    cmriverside Posts: 32,186 Member
    sijomial wrote: »
    kshama2001 wrote: »
    sijomial wrote: »
    kshama2001 wrote: »
    sijomial wrote: »
    Some of the exercise estimates used by MyFitnessPal source the Compoendium of Physical Activities.
    And that gives estimates in METS which as the OP correctly points out is inflated by 1 MET - a gross calorie estimate not a net calorie estimate.

    For short duration exercise it's a trivial difference.
    For long duration exercise it starts to mount up - it's a particularly large percentage deviation for long duration and low rate of burn exercises such as walking.

    Yes the cycling speed categories are badly inflated for any sort of decent road bike - but I don't know any serious cyclists who would use such a dreadful way to estimate calories when much better ones are commonly used!
    There's a world of difference between a mountain bike being riden in tough terrain (or over mountains!) and a fast road bike on the flat which speed ranges don't take into account.

    BTW - please can people stop insisting the exercise estimates were accurate when they mean their weight loss results were on track. That verifies your calorie balance but it certainly doesn't validate all the elements involved on both sides of the CICO balance. Unless you have validated your exercise against a known good method (such as a power meter on a bike) you simply know the estimates worked for you - not that they were accurate.

    Did someone say they were accurate? I guess I didn't read any of the responses that way.

    Same goes for the tracking devices. All are standard algorithms that are used as estimates. Not really accurate, just close enough.

    I said the particular walking entry I use was accurate for me (Walking, 3.0 mph, mod. pace, walking dog)

    Because I eat a lot of whole foods and get my syntax for those entries from the USDA database, I am very confident in my food logging but was having trouble losing weight until I changed some exercise choices.

    I gave strength training examples above.

    For gardening, I give myself credit in a range from no credit to full credit. If there's a shovel involved, I take full credit. If I'm clipping roses like Annette Bening in "American Beauty", I take no credit. (However, if I am pruning up a storm, that's different.)


    Silly picture removed.

    Same question - how did you validate that the walking estimate was accurate for you?
    Hooked up to a breath analyser in a gym perhaps to measure oxygen uptake?

    How did you validate your BMR was accurate?
    How did you validate your activity multiplier was accurate?
    How did you measure your TEF?

    Again getting the right results doesn't not validate all the elements involved.

    e.g.
    When I started out I used inflated exercise estimates from a HRM and gym machines.
    Adjusted my daily goal and proceeded to lose 1lb a week as predicted - my exercise estimates were still inaccurate, losing exactly on goal didn't change that, I just compensated for the inaccuracy.

    Losing as expected over time is all the validation I need that I am using good entries.

    I'm afraid people will read your post and think weight loss is a big complicated technical process, when our bodies can handle these processes. We just need to burn more than we eat. We don't need to be perfectly, excruciatingly accurate for this to happen.

    I would hope they would realise that reasonable estimates in all aspects and the common sense to make adjustments based on results is what's actually required. Not chasing perfection or absolute accuracy.

    Look how simple my example was. Logged my food completely but not absolutely accurately, used easily available exercise estimates, adjusted my calorie goal based on actual weight lost over time = success.

    For me as a high volume exerciser (like the OP appears to be) investing some time in making my exercise estimates fairly accurate made sense - for a low volume exerciser it's a small part of the puzzle.

    We're all saying exactly that.

    Except OP, who wants the numbers spit out on this site to fit every person.

    They fit me. In the last ten posts, they fit kshama and lgfrie. Both of them are dialed-in on their food logging, and so am I, so it makes sense to just keep working the numbers until OP makes them fit, but not make a blanket statement like he did in the first post, that there is a, "Serious flaw with Myfitnesspal." There is some flaw in the way he's using the tool, but the tool itself spits out numbers that work for some people and not for others. No way to really know what the mistake being made really is. It could be his logging, his height/weight, his discrepancy in the MPH/time, outlier exercise intensity, or some other input error, or something not mentioned yet.

    That is why I made a contradiction in the first reply and used myself as an example. :flowerforyou:
  • jhanleybrown
    jhanleybrown Posts: 239 Member
    edited September 2019
    sijomial wrote: »
    Not sure what you are implying but whatever. Pretty sure I can get ave soloed from a cyclometer and Mike age from a map...

    To those who exercise a lot as part of your program:
    - you can't use calorie estimates on this app...they are all *really* high.

    You have a valid point about the programming flaw but "they are all really high" is an exageration.
    Gross v. Net does not make everything really high, it might make it maybe 100 cals too high an hour for me as an average sized man. Big percentage difference for walking, much smaller difference for an hour of intense cardio.

    You also can't extrapolate the cycling speed ranges which I also know are badly inflated for me and my power output to all of the hundreds of other exercise selections in the database.

    Give the strength training entry a go and although it's impossible to measure outside a sports science lab I bet you won't think it's "really high" if you are moving a decent amount of weight in that hour.

    I don't know....every exercise I do then research...its coming out 30%+ high. I'm not a weight lifter and not really interested in that...but biking, running, swimming, walking, yoga (I'm a former rower and new sculling again bit haven't looked that one up...)...

    *all* at least 30% high on calorie output.

    For all you MFP white knights: I still like the app! I've been using it for several years. I'm not advocating abandoning it. Also, I've researched my most common foods and have found that on the food side, it's probably the most accurate systematically. So I'm not anti-MFP.

    I have no involvement with this guy but I think he's onto something about this app over estimating burn:
    https://lluniversity.com/why-net-calories-suck-and-your-my-fitness-pal-app-is-failing-you/

    For the record: I like this app. *But* it appears to systematically over estimate exercise calorie burn by at least 30%. Which affects some more than others based on your mix of calorie restriction vs exercise for getting to a deficit.
  • heybales
    heybales Posts: 18,830 Member
    Another reason why exercise calories can "appear" to overestimate, perhaps greatly.

    And it's beyond the Net / Gross comparison.

    MFP already has an estimate for your per min burn, not based on BMR, but rather BMR x 1.25, 1.4, ect. Activity Level.

    That is the existing burn rate the exercise database entry which contains BMR is then being added too.
    So if you selected higher than sedentary, the effect is even worse.

    And as pointed out, low calorie exercise burners done for long periods of time just compound the negative effect.

    Yes MFP could fix this easy enough since they have both pieces of info.

    But then it would have to be modified when an activity tracker is synced - since those are all replace type systems on their end, rather than MFP's add type system, the effect works correctly, even when a workout is synced over.

    Because really the database by itself reflecting what is burned in a chunk of time - which would obviously include your base burn - is based on studies, and from that for example the common accurate formulas for walking and running.
    Only problem with walking/running is the wide ranges of the intensity (pace) compared to say a formula taking exact distance and time.

    I've even compared the biking entry to watts for a general non-steep non-windy up and down route, when I compared the watt burn and added in the non-moving time like at lights for BMR (though obviously burning more than BMR right then).
    When I used my distance and total time on the bike (not just moving time) and I was actually in the middle of the speed range of the entry - it was within 5%.

    Swimming has 2 entries, and you'd think that is bad - but if you take another good swimming estimator and throw in some normal human fast and slow times (not olympic level) - you discover the difference between those extremes isn't even that great, so 2 entries more than covers the potential.

    I'd be curious what sources you are using that gave the 30%.

    Now, you garden for 4 hrs and log it - ya you'll have problems since your accounted for burn may be 110/hr and it's logging 140/hr in addition to it, rather than the 30/hr more it should be.