Weight loss as a math problem, but which numbers are wrong?

Options
13

Replies

  • WalkingAlong
    WalkingAlong Posts: 4,926 Member
    Options
    Mr_Knight wrote: »
    WalkingAlong wrote: »
    Fitbits are totally designed for NEAT.

    You know the Scooby TDEE calculator? And the one at IIFYM?

    That's what is inside a FitBit.

    All a FitBit does is use the motion sensor to guess where you are on the Sedentary -> Active scale, and then guess at your calorie burn based on the usual TDEE math you can find at any web site.
    Not exactly, but it does use population averages. The accelerometer compares the motion pattern it detects with motions of various walking/running test databases of people of a similar profile (gender, height, size, etc.) Say it finds that at that moment you appear to be walking 3.8mph, based on your movement. It takes the METS value for that activity from the public compendium and multiplies it by your BMR estimate, for that moment, to get a calorie estimate for that moment. It does that all day and sums them up. So it's a little more sophisticated than saying "low step count = sedentary = 1.2 times BMR".

  • StaciMarie1974
    StaciMarie1974 Posts: 4,138 Member
    Options
    I assume trackers are off by 5-10%. They just can't be 100% accurate.

    Personally I know now my Fitbit overestimates my burn. Why? Because I'm trying to maintain and have continued to lose weight while eating at/just below my Fitbit burned #. But that will vary by person. And even in our best efforts we can't be 100% accurate on food. So between calorie burn overestimation, consumed underestimation, and potential water weight - you're clearly moving in the right direction.

    Continue to focus on the math/science if that's what you need to do. But in the end, if you're weight is less today and your fitness level higher now than 30 days ago: Good job!
  • mburgess458
    mburgess458 Posts: 480 Member
    Options
    To answer the initial question of which number is wrong, they are all somewhat wrong. The TDEE formulas are estimates so they are definitely not 100% accurate. Fitness trackers are estimates so also not 100% accurate. Your estimation of calories consumed is just that, an estimate so also not 100% accurate. And your calculation of weight loss is affected by how hydrated you were at your initial and final weigh in.

    It is fine to look at it as a math problem, but it really is more like a science experiment. Each month all of the errors are going to be different. Keep compiling the data and you'll get closer and closer to knowing your "true" TDEE, calorie burns, calories consumed, and relative hydration level. The other problem is that as you're compiling the data you should also be losing weight so the true value of everything will keep moving also.
  • leggup
    leggup Posts: 2,942 Member
    edited February 2015
    Options
    subversive99 wrote: »
    WalkingAlong wrote: »
    OP- I did the math over 6+ months and my Fitbit did a good job of predicting my weight loss, over 35 lbs. You could fit the averages less well though, or track food less well. Knowing how much you vary from the predicted is all you need, though. If you want to lose twice as fast as you are, you know how much to reduce your intake by, for example.

    Good stuff, thanks. Funny enough, since I posted this thread two days ago, I lost 1.5 pounds Sunday morning and another 2 pounds this morning. The body is a complicated machine all right...

    This was what I was going to talk about. That "5 lb actual loss" - was it a weight that you weighed in at only one day? As in, do you weigh in once a week or once a day? Weight fluctuations can be pretty wild, especially if you eat more on the weekends, if what you eat contains alcohol or more sodium than normal.

    Weight loss is not linear. When I was set to lose 1-2 lbs a week I could see fluctuations of up to 6 lbs in the same week. I entered my weight every day and saw the downward trend, so I didn't care about any single weigh in.
  • subversive99
    subversive99 Posts: 273 Member
    Options
    britishbroccoli wrote: »
    This was what I was going to talk about. That "5 lb actual loss" - was it a weight that you weighed in at only one day? As in, do you weigh in once a week or once a day?.

    I usually step on the scale every morning but I only record my weight and BF% once a week so I'm certainly familiar with fluctuations. The idea with weekly weigh in on the same day each week is to gradually eliminate the effect of them.
  • WalkingAlong
    WalkingAlong Posts: 4,926 Member
    Options
    Since you use a Fitbit, consider also using Trendweight. Log your weight daily at Fitbit (or get a wifi scale) and watch your 'trend weight' on the Trendweight site. It removes the noise and lets you see a more refined actual trend.
  • bwogilvie
    bwogilvie Posts: 2,130 Member
    Options
    WalkingAlong wrote: »
    Since you use a Fitbit, consider also using Trendweight. Log your weight daily at Fitbit (or get a wifi scale) and watch your 'trend weight' on the Trendweight site. It removes the noise and lets you see a more refined actual trend.

    This is a good idea. I have done that for two years, first with an Excel spreadsheet and now with TrendWeight.com, and it has been very helpful.

    OP, you might be interested in the results of one study that showed that laypeople typically underestimated their calorie intake by over 400 calories a day, and even trained dietitians missed around 200 calories a day.

    When I started, I was missing around 200 calories a day somewhere, probably a combination of inaccurate food logging and overestimating energy burn. I've now tweaked things so that my logging "works," in that the numbers I log are fairly close to my actual weight loss/maintenance, but that has involved setting my activity level to sedentary (despite a fair amount of time on my feet)—suggesting either that my BMR is near the low end of the range or that I'm still missing some calories. I'm not worried about it, though; as long as I continue to be consistent, I'll do fine.
  • Cheyenne_K_
    Cheyenne_K_ Posts: 31 Member
    Options
    I only read the first page of comments so sorry if this has been asked/discussed already.... so my polar heart monitor says I burn about 3000 calories a day. Factoring in a 12% margin of error, that puts me at 2640. I usually eat between 1800-2000 calories a day. Let's add 200 calories to say margin of error for tracking to my higher calorie day. 2640-2200= a deficit of 440. That should be almost a pound a week weight loss. Based off of Slinfold tests, pictures, and scale my weight has not changed.... WTH?
  • sofaking6
    sofaking6 Posts: 4,589 Member
    Options
    subversive99 wrote: »
    WalkingAlong wrote: »
    OP- I did the math over 6+ months and my Fitbit did a good job of predicting my weight loss, over 35 lbs. You could fit the averages less well though, or track food less well. Knowing how much you vary from the predicted is all you need, though. If you want to lose twice as fast as you are, you know how much to reduce your intake by, for example.

    Good stuff, thanks. Funny enough, since I posted this thread two days ago, I lost 1.5 pounds Sunday morning and another 2 pounds this morning. The body is a complicated machine all right...

    I was gonna post - yes it takes 3500 calories to make a pound, but your body still has to process all of that food and turn it into energy and person, or vice versa. That isn't going to happen at the same exact rate 24/7.
  • Mr_Knight
    Mr_Knight Posts: 9,532 Member
    Options
    Cheyenne_K_ wrote: »
    ....so my polar heart monitor says I burn about 3000 calories a day. Factoring in a 12% margin of error...

    Your margin of error is WAY higher than that, because you are using your HRM completely incorrectly.

  • thatstheticket
    thatstheticket Posts: 16 Member
    Options
    mburgess458 wrote: »
    If he had the same tendency for water retention when he began his weight loss journey as he does now, then this extra weight has already been incorporated and can be ignored.

    For example:
    Starting weight: 180lbs + 5lbs water retention = 185
    Ending weight: 170lbs + 5lbs water retention = 175

    Net loss is still 10lbs.

    This would make sense if water retention was constant over time, which it is certainly not.

    If I have something like hot and sour soup or a bunch of pickles (which aren't bad in terms of calories but can have tons of sodium) I will gain about 3-4 lbs.....for a few days. It's just water retention. If he doesn't eat the exact same amount of sodium each day his water retention will go up and down by at least a few lbs.

    That's why I weigh myself daily and focus on the 7 day average weight. It will still have some random water weight fluctuations but averaging over a week smooths things out a lot.

    True! What you say makes sense. >:)

    In this case it is more of a simplified situation, similar to what happens with me. My "water weight" still fluctuates the same +/- 2lbs and this hasn't changed after losing weight.

    I just ignore the gains and focus on the losses!! It does seem to average itself out like you say and has worked for me so far
  • subversive99
    subversive99 Posts: 273 Member
    Options
    So, thought I'd post my progress after 4 more weeks, so those following along can see the update. Also maybe this will help some analytically inclined folks to sort out their own numbers.

    Second 4 weeks data (Feb 7 - Mar 6, 2015)

    Eaten 68401
    Burned 95106
    Expected Loss -7.63
    Actual Loss -4.1

    Total Data for 8 Weeks (Jan 10 - Mar 6, 2015)

    Eaten - 127656 (2279/day avg)
    Burned - 190036 (3393/day avg)
    Deficit - 62380 (1113/day avg)
    Total Expected Loss - 17.82lb
    Total Actual Loss - 9.10 lb

    So, what am I taking from this? Well, if I play around with numbers I can see that if I assume my TDEE is being overestimated by 10% and my calorie counting is being underestimated by 10%, the numbers match up almost exactly to the actual loss.

    In reality, I will try to focus on sticking to the 2000 calorie/day goal I have set for myself a bit more strictly. If I could bring that average down from 2279/day to 2000/day, that would result in an extra 3.2 pounds lost per 8 weeks, shortening my path to maintenance by a non-trivial amount.
  • Farm_Girl_Strong
    Farm_Girl_Strong Posts: 81 Member
    Options
    Just wanted to add, that I wear a Mifit Flash right now, no though for me it's more. Reminder to get up and do things.... Today my whole family has the flu, and I spent 1.5hours rocking my 11month old who is miserable and the Misfit tracked me doing 3000steps... I laughed when I saw that!

    So as great as trackers can be, they don't know the difference between me walking and the motion from the chair. So I always assume my tracker will over estimate by about 15% each day. (Today obviously being more)
  • segacs
    segacs Posts: 4,599 Member
    Options
    subversive99 wrote: »
    Well, if I play around with numbers I can see that if I assume my TDEE is being overestimated by 10% and my calorie counting is being underestimated by 10%, the numbers match up almost exactly to the actual loss.

    Sounds reasonable. In fact, I'd assume that, for most people outside of a laboratory setting, it would be near-impossible *not* to underestimate calorie counting by 10-15%.

    The absolute numbers don't matter. As long as you're consistently off by about the same amount, it should work on the directional numbers.
  • subversive99
    subversive99 Posts: 273 Member
    Options
    Little update here. Progress continues. Making it a math problem is still working for me. Been through 4 x 4 week spans now. Here are the total numbers.

    Weeks 1-4
    Eaten: 59255
    Burned: 94930
    Expected Loss: 10.19
    Actual Loss: 5

    Weeks 5-8
    Eaten: 68401
    Burned: 95106
    Expected Loss: 7.63
    Actual Loss: 4.1

    Weeks 9-12 (nearly 3 week stall in here...)
    Eaten: 75085
    Burned: 97578
    Expected Loss: 6.43
    Actual Loss: 1.6

    Weeks 13-16
    Eaten: 72671
    Burned: 96805
    Expected Loss: 6.9
    Actual Loss: 4.9

    Total for 16 weeks:
    Eaten: 275412
    Burned: 384419
    Expected: 31.14
    Actual: 15.6

    So, am I upset that my actual loss doesn't match my expected loss? Not at all. I now know that I can safely assume 10% under counting on my food tracking, and 10% over estimation on my burn rate from fitbit. I am sitting at almost exactly 1 pound per week of actual loss, which is a great and sustainable rate, while eating a lot of delicious food.

    See below for my trendweight chart to date.


    yIbV37c.png
  • girlviernes
    girlviernes Posts: 2,402 Member
    Options
    Very cool. Thanks for sharing your data :)

    Another thing to keep in mind, not sure if this was mentioned, is that TDEE is always an estimate in the first place, so overtime if your logging is (relatively accurate) you will get a better idea of your actual TDEE.
  • maidentl
    maidentl Posts: 3,203 Member
    Options
    Cheyenne_K_ wrote: »
    I only read the first page of comments so sorry if this has been asked/discussed already.... so my polar heart monitor says I burn about 3000 calories a day. Factoring in a 12% margin of error, that puts me at 2640. I usually eat between 1800-2000 calories a day. Let's add 200 calories to say margin of error for tracking to my higher calorie day. 2640-2200= a deficit of 440. That should be almost a pound a week weight loss. Based off of Slinfold tests, pictures, and scale my weight has not changed.... WTH?

    A HRM is not designed to be worn all day. You're eating too much.
  • Angel_Grove_
    Angel_Grove_ Posts: 205 Member
    edited May 2015
    Options
    lynn_glenmont wrote: »
    I've done this. Actually I've done this twice, over 8 weeks in the fall of 2013, shortly after starting MFP, and over 9 weeks in the fall of 2014. But I did it to calculate my NEET (the amount of exercise I get from week to week is way too variable to go with TDEE), and got a number more than 200 calories higher than MFP calculated in 2013, and a number about 400 calories higher than MFP calculated in 2014. It didn't send me down some rabbit hole trying to figure out what I was doing wrong. I did a mental happy dance -- more food if I want it, more weight loss if I don't. No need to try to stick to the miserable 1460 calories MFP originally suggested.

    Instead of assuming the TDEE calculators are right, why not just assume your logging of calories consumed is correct, add the 59,255 calories you ate over four weeks, the calorie equivalent of the five pounds you lost (17,550), and divide by 28 to find your "real" TDEE of 2740. Even if the TDEE calculators are right and there's something wrong with your logging, isn't it a whole lot simpler to use the TDEE that reflects the way you log, rather than to try to figure out where the mysterious problem(s) is/are in your logging, and then try to find a way to correct them? Your TDEE is 2740 "OP calories," where "OP calories" are the units in which the OP measures the energy in food and the energy his body uses.

    I went with 8 weeks of data because I'm a woman, and I wanted to smooth out the hormonal fluctuations. I would think 4 weeks of data would be enough for a man, but, hey, in another 4 weeks you can toss two months of data together and see what you get.

    Best of luck. Embrace the numbers.

    I'm pretty sure this is the best advice I've seen on MFP to date, and I'm kind of pissed I didn't come up with it.
  • segacs
    segacs Posts: 4,599 Member
    Options
    subversive99 wrote: »
    So, am I upset that my actual loss doesn't match my expected loss? Not at all. I now know that I can safely assume 10% under counting on my food tracking, and 10% over estimation on my burn rate from fitbit. I am sitting at almost exactly 1 pound per week of actual loss, which is a great and sustainable rate, while eating a lot of delicious food.

    WINNING!

Categories