Effect of Activity Level on my TDEE

FIT_Goat

Warning, this is a long post. It's full of rambling and talks about data. There are a couple graphs just to break up the monotony. Right now, I can't draw any conclusions based on what I am doing. This is the basis for an experiment, on myself, that I started last week. I have been tracking everything for a while. I am a nerd. I track every number I can think of. I track my macros, I track my daily weight, I track my ketogenic ratio, I track my pulse, etc. You can skip this if it's not your thing.

One thing, in particular, that I am interested in tracking is my actual TDEE. The formulas are great, but they are all estimations based on other people. I can get a better estimation tailored for myself based on how many calories I eat and how that effects my weight. I also want to track my personal TDEE to see if I notice a large decrease in it, which would signify that my body is adjusting to my long-term diet in a negative way.

What I do, is I take a 30-Day TDEE estimation. I calculate how many calories I ate over a 30 day period, and then I look at the change in my trend-line for the 30 days relative to that intake. The formula is pretty simple (Calories_eaten - (3500 * (end_weight - start_weight))/30. Basically, a pound of weight change represents 3,500 calories in excess or deficit compared what what my body used.

Each day, I drop the oldest data point and add the next one. So there is a rolling 30-day average estimation for my TDEE.

The values that I calculate range from 2,392.55 up to 2,817.78.

I have assumed, up to this point, that the reason for any variation comes from the most obvious change in calories used, Exercise! Basically, those times when my TDEE was higher were times where I was "extra-active" on average for the 30 days and the times when the TDEE was lowest was when I was not quite as active, on average for the 30 days.

This meant, I could probably find a base-line TDEE if I could find a way to account for the level of activity. Well, I actually have a proxy that I can use to measure my relative activity. I wear my FitBit almost 24/7. I don't pay much attention to it, but I wear it all day and when I sleep. The only time I am not wearing it is when I shower, it is charging (I am usually watching TV anyway), or if I go swimming. I did take it off when I participated in a mud-run, for example, but that is one day out of the entire time.

So, I put in the FitBit's estimated calories for each day. Then I took a 30 day rolling average. And, I used those numbers to create an adjusted TDEE. Basically, the higher my FitBit calories (relative to the lowest number), the more calories that I took off my TDEE calculation. I scaled it because my FitBit calories ranged from 2,509.2-3,142.8 ... a range of 633.6 calories between active periods and inactive ones. My TDEE only ranged by 452.22 calories. So, a time when I averaged 3,142 calories on my FitBit would take 452 calories off my calculated TDEE for that time. A time when my FitBit averaged 2,510 calories would take almost nothing off my calculated TDEE.

The end result of all this work should be a fairly narrow range of values that represent my "sedentary" TDEE level... basically, the lowest level I actually tend to hit over time. Every time I average higher activity, it would cause my TDEE to rise above that level.

I did these calculations on my spreadsheet and I graphed them. What did I expect to see? I expected my FitBit TDEE to have basically the same shape as my calculated TDEE. I knew it wouldn't be perfect, but I expected it to be pretty close. It would be higher, naturally, because it seemed to estimate high. Below the FitBit and TDEE lines, I expected to see a fairly flat "Adjusted TDEE" line. Those times when I was most active would have been shaved down to the level when I was least active.

This is what I wanted to see:


This is not what I saw. What I actually saw was:


Let's put them together because it's neat looking:


My adjusted TDEE line actually has a wider range than my original one (476.5 calories) and is not flatter than the original. The amount of exercise does not seem to be effecting my TDEE. Yeah, the TDEEs are both higher near the beginning (but I was also 30 pounds heavier--which alone can account for a 200 or higher calorie drop).

I then decided to graph my calories consumed (scaled to align with the highs and lows) to my TDEE and see if maybe how much I ate was effecting how much I burned. Maybe, when I eat more, I burn more. That one doesn't look much better.



Finally, I decided to graph my Ketogenic Ratio (scaled to align with highs and lows) to my TDEE and see if it matched. This is basically a ratio that compares the relative amounts of carbs/protein/fat and produces a number. Basically, carbs cause it to go down quickly, protein causes a slight decrease, and fat causes it to go up. It is supposed to be 1.5 or higher. My 30-day average ranges from 1.76-1.94 (although I am increasing that upper end right now). Shockingly, the shape of the graph was very close. And when I shifted it over a few days it aligned very closely. This is that graph.



I am going to continue graphing while trying to increase my average ratio... and see if my calculated TDEE follows suit or if it breaks off. It could very well just be coincidence at this point.

I also track my calories compared to my rate of loss. Those also seem linked (I can show that another time), except recently. As I have consciously been raising my calories and my ratio, my loss has been slowly accelerating and the calorie and loss-rate lines are diverging. It could be temporary (has happened before), but we'll see. It could be interesting to see what happens there.

tru2one

Thank you, Professor Frob! ;-)

Wow...just...wow. Thanks so much for taking the time to track and analyze this stuff! I think that there's much to be interpreted by your findings, but one of the takeaways I notice is it seems you're basically proving the effect of being in a ketogenic state on your TDEE, ie: higher ketogenic ratio drives TDEE upward? Maybe more so then physical energy expenditure?

I could be way off base...science was never a strong subject for me! lol What are your initial conclusions from the data presented (realizing it's still in hypothesis stage and more will be revealed as you continue to track)?

FIT_Goat

My initial reaction, which is based on the reason I started tracking this, is relief to see that my calculated TDEE seems to be minimally effected by the amount of weight I have lost. The last time I lost a lot of weight on a low-calorie "balanced" diet. I used MFP and based on their recommendations for my height/weight/age and my desired rate of loss (the maximum, of course), I ate 1,200-1,400 calories a day. I went from 97.5kg to 73.6kg (from 215 lbs to 162 lbs) over the course of a year! Success! Except that I ended up with a maintenance amount of 1,600 calories. If I ate much above that, I would slowly gain weight. I am only 5'4" (163cm), but that was still about 15% below what my BMRxSedentary calculations said I should be at. I was starving all the time, and eventually stopped tracking and let most of that weight go back on over the course of a few years. I started at 92.5kg this time (203.5 lbs) and am down to 75.5kg (166.1 lbs). So, basically the same amount of weight lost. Except, this time I've done it while averaging 1,900 calories/day and managed to keep my TDEE around where I started.

So, that's my current big takeaway. This way of eating has allowed me to drop a lot of weight, at a decent rate, while remaining metabolically stable. I'm also not hungry or exhausted all the time, but that's not really something I can show in the charts. I also track an estimate for my BF%, and while I appear to have lost some lean body mass, it is minimal. It is also possibly within the margins of error based on how poor my actual BF% estimates are and were. I believe eating this way has worked to preserve more muscle mass than my first weight loss attempt.

If the data continues in the way it looks like it has been, it would seem to show that the higher k. ratio I maintain, the higher my metabolism. Activity level seems loosely, if at all, related to how much I end up actually burning. Probably because I subconsciously compensate for that activity by reducing small movements that I may not be aware of. I can't say why it doesn't effect my caloric burn as much as it should.

But, yeah... if I drive my ratio way up... and my TDEE peaks along with it. That's a good indicator that there's something going on with metabolism and macro-nutrient composition.

tru2one

So, just curious...do you remember what your BF% was the first time around with low cal when you hit goal? I would guess it was higher than this time and that there was significant lean muscle loss. Which would totally explain why TDEE is not affected as much this time, since lean muscle tissue expends more energy...right?

FIT_Goat

So, just curious...do you remember what your BF% was the first time around with low cal when you hit goal? I would guess it was higher than this time and that there was significant lean muscle loss. Which would totally explain why TDEE is not affected as much this time, since lean muscle tissue expends more energy...right?

Yeah. That would be a reason. Lean body mass burns more calories. So, I could have been a lower weight, but had proportionately less muscle mass.

There's also Adaptive Thermogenesis, which also plays a role. Basically, after prolonged periods of dieting, your body has down-regulated to some degree and you're burning fewer calories than someone with an identical composition, weight, and age would burn if they had always been that weight. So, I think some of it also has to do with that. Especially since I was at such low-calories for so long.

Unfortunately, I wasn't doing things smartly that time. I have no idea what my BF% was at the end, or start (both would have been nice to see which diet really spared more lean body mass). There's no way to know for sure. But, it was a primary concern when I decided to lose weight this time. I didn't want to go down that road again.

Leonidas_meets_Spartacus

There are lot of factors and it would be hard to say with just weight lost and considering your BF% changes, BMR or RMR measured. Great effort though.

FIT_Goat

Taking changes in body fat, bmr, and rmr into account, I expected the adjusted TDEE line to have a negative slope. All of those things should be working to lower my energy expenditure.

The fitbit does some compensation for this when I enter my weight. It reduces my assumed caloric burn to my new calculated bmr.

In any case, I know there's not enough control here to draw conclusions. There is just enough to drive further self experimentation. Even then, the results would not be sure and could potentially only apply to me and not others.

I still find it interesting. I wonder what would happen if I used my calculated change in bmr (both from total mass and lbm formulas) to further adjust things. I didn't do it originally because the fitbit includes a similar calculation. It wouldn't improve things, since it would drag the left side farther down, but it could be interesting.

Leonidas_meets_Spartacus

Try including body fat measurements and also get vo2, metabolic tests to actually measure your calorie burn. Fitbit or HRM have no clue what you are burning. HRM for me was off by more than 20% on calorie expenditure and burn rate at various HR. You don't need to do every day, but once you know the base line burn rates, you can estimate from your HR zone. Also, my Body fat loss was a steeper curve than the weight loss curve, because weight is dependent on lean muscle mass and ofcourse water weight is a big factor. You are doing awesome and see if it's possible to tweak with additional data like actual burn rates and BF loss.

FIT_Goat

It is pretty obvious that my FitBit doesn't have any idea of how much I am burning. That's why I can't find a strong relationship between the two. it is certainly likely that my activity level bears some relationship to my calories burned. I am much more confident in my TDEE than the one produced by the FitBit (for reasons I explain below). What I attempted to do was to find a way to relate the amount of calories my FitBit thought I was burning to the actual amount I was burning. In theory, even if the fitbit is off by 20%, the times when it thinks I burn the most calories should align with the times I actually do burn the most calories. And the times when it thinks I burn the fewest calories should align with the times I actually do burn the fewest. Did I expect perfect alignment between the peaks and valleys and the degree of difference? Not really. But, I expected some relationship. My FitBit may not be able to accurately tell how many calories I burn walking the dogs, but it knows I was walking around for an hour instead of watching TV.

I did go and check to see how my weight loss (as well as the fact that I am older now) should have effected my RMR/BMR.

* Changes in LBM would result in a 51 calorie drop in BMR (Sterling-Pasmore Equation)
* Changes in total weight result in a 166 calorie drop (Mifflin Equation (RMR)) to a 228 calorie drop (Harris-Benedict equation).

So, according to those results, my adjusted line should be dropping between 51 and 228 calories over the time measured on these charts. Probably even a little bit more because those both would be multiplied by activity level to some degree (since I never rest 24/7 for 30 days in a row).

My calculated TDEE isn't based on any device that measures movement or heart-rate. It is based on a weighted trend-line and my record of calories consumed. My trend-line is reasonably accurate, especially for comparing change over time. The calories are based on weighing everything I eat. The calories probably have the largest possible error. We can say they are off by +/- 10%. Over a 30 day average, that error approaches +/- 35 calories/day ( sqrt( day1_err^2 + day2_err^2 + ... + day30_err^2 ) /30 ). Even if my calories have a 20% error rate, it amounts to +/- 70 calories/day. So, there is a potential source of error. But, if we assume that the error is random (which I think is safe to assume), then over 30 days it tends to minimize itself to an amount we can consider reasonable.

For these reasons, I feel that my calculated TDEE is reasonably accurate to a degree of +/- 75 calories. And that is being pretty pessimistic. I looked at the variance in my trend-line. The maximum daily variance was 0.25 kg. This amounts to less than +/- 7 calories a day over a 30 day period in error. But, let's be super-safe and up that to 1.1 kg in variance. That's +/- 30 calories a day in possible error. If my calorie counts have an error of 10%, that adds another +/- 35 calories a day in error. For a total of +/- 65 calories a day. We would have to take both pessimistic assumptions to approach +/- 100 calories in error/day.

TDEE = BMR + Activity + Thermic Effect of Food.

While I can be fairly confident about my TDEE, I can't be sure about the weight of each of the components that make it up. My BMR could be going down, while my Activity level has risen slightly (in ways my FitBit doesn't notice), and the Thermic Effect of my Food could also be going up.

Anyway, the short answer here is that I do have a reasonable picture of what my actual total burn rate is. While, I don't know the weights of the individual components that make up my burn rate. In the future, when I near my goal weight, I want to get a complete workup done. One that actually gives me an [more] accurate measurement of my body fat and actual metabolic rate. Of course, even if I have my real BMR/RMR, then I need to determine an accurate multiplier for my activity level. That's guesswork too. In fact, I would probably use my calculated TDEE to help set it. That's actually how I go about setting things when I use the keto calculator. I need to use a moderate activity level to bring their estimate of my BMR up to my actual TDEE.

Also: All of these data points were created using data one month or older into the diet (to reduce the effect of initial water weight losses and reduce the variance as much as possible).

FIT_Goat

Leonidas, I just glanced at your profile (you referenced it in a post in the keto group) and saw that you've had an 18% increase in metabolic rate while dropping a ton of fat. Very interesting.

That definitely makes me want to actually get measurements done. More, it makes me regret not having baseline measurements to compare to.

Leonidas_meets_Spartacus

Better late than never, you are doing great job with numbers. You will do great if you can add some important stuff like BF and metabolic stuff to draw some conclusions.

Nicole9187

I just wanted to say I really enjoyed reading all this information!