Weight Training: Cut Edition

Previously: One / Two / Three

In my last post on the topic, I said I was going to go on a cut. Well, now the cut is done, let's discuss a bit!

Was it a success? Well my stated goal was to acquire a six-pack, so in that sense, no. But I learned a ton, and I did visibly lose a considerable amount of fat and slimmed down from 90 kg to 83 kg over 15 weeks.

I scrupulously transcribed my daily caloric expenditure and intake in a spreadsheet. All the numbers and metrics I discuss in this article are recorded there!


Initially, I went for a 700 kcal daily deficit. This would add up to a 4900 kcal deficit per week, which represents about 630 grams of fat. As it turns out, this is exactly 0.7% of my starting bodyweight of 90 kg, which is the recommended weekly loss.

I did struggle with the strict regimen initially. This was not an issue of will, but rather an issue of organization. Too often I accidentally overshot my goal — in fact my average daily deficit in the first four weeks was a measly 385. Oops.

A remark on my method here: while I'm a fan of using a Fitbit smartwatch to track caloric expenditure, it does add a bit of logistical uncertainty, as you must estimate what the expenditure will be by the end of the day. Mostly, I adapted to cope with this by having very light dinner meals, and potentially a snack before going to bed. I'll have more to say about the Fitbit towards the end of the article.

After four weeks, I ran a refeed week, where I ate at maintenance. This is done to prevent metabolic adaptation in which your body gets used to get less calories and adapts to burn less of them. That week was also a deload week where I decreased my weights at the gym (otherwise I kept training using the previously outlined program).

In the second four-week period, I was able to average between 600 and 700 kcal of deficit daily - yay!

Is It Working?

So we have our caloric deficit down. But is it actually working as expected?

To do this I compared my recorded weight to two measures of the estimated weight given the recorded caloric deficit. The first estimated weight is global: it's my starting weight minus the fat weight loss predicted from the cumulated caloric deficit. The second estimated weight is local: it's computed in the same way but only for one week, starting from the starting weight just before that week.

In order to predict the fat loss from a caloric deficit, you simply need to know that a kilogram of fat is equivalent to about 7700 kcal (or 3500 kcal to a pound).

To ascertain whether I was on track, I further computed two weekly metrics: the average local error is the difference between the average locally predicted weight and my average weight for that week ; the average global error is the same, but using the average globally predicted weight instead.

Taking these averages allows smoothing over the noise inherent to weight measurements: there can be big variations in water weight from one day to the other (even though I always weighed myself in the morning after going to the bathroom and before drinking anything).

Another thing I did to be able to track my progress was compute a smoothed weight measurement: the daily smoothed weight is the average of the the weight in a 5 day window surrounding the day under consideration. This was also the basis I used to compute the local weight estimation.

After the initial 5 week period (4 weeks + refeed week), I had lost 1.3 kg. That is slighty (about 200 g) more than the model predicted. In general, the average local and global errors were small (<= 100 g, either above or under) during this period.

After the second 5 week period, I had lost 3.6 kg in total, a whole kilogram more than the global prediction! The data was much more noisy here, mostly the global error increases in the negatives (underestimation) but it shot back towards zero because of a weight measurement bump that normalized quickly. The local error varied between 0 and -500g (underestimation).

There is only one remaining uncertainty: was this weight loss fat loss... or muscle loss?

This is quite tricky to measure. The impedance scale I bought isn't worth shit. Methods requiring a fat caliper are janky... They require multiple measurements, sometimes in location you can't access on your own (back of the arm, shoulder blade). I bought a cheap one and it's also not quite clear how much you need to press.

The impedance scale at my gym (Basic Fit) seems a bit more accurate though, reporting 17% body fat somewhere within the second four-week period, and 14% at the end of the cut.

In my spreadsheet, I estimated my starting percentage at 18% (based on visual comparison to some online reference), which — assuming 100% of weight loss is fat loss — would put me at 16.7% at the end of the four weeks period and 11.2% at the end of the cut.

There is another method, which consists of taking measurements of your waist and your neck (using a regular meter), and which is reportedly relatively accurate — you can find a calculator here.

All the research points to the fact that, given a moderate deficit (the figure of 0.7% of your starting bodyweight is often given), the overwhelming majority of the weight loss should be fat — if you keep doing resistance (weight) training and intaking enough proteins, which I did.

There is a theory that you can only mobilize so much calories from fat mass daily. Let's look into it.

A Detour: How Fast Can You Actually Lose Fat?

There is a theory saying that there is a limit on the calories that can be made available from fat over a given time period. However, it's not quite clear what that limit is.

A 2005 study in theoretical biology found it to be 31 kcal per pound of body fat daily (about 68 kcal per kilo). However, the author reportedly later developped a better model that brought the number down to about 22 kcal per pound (48.5 kcal per kilo).

There are a couple of caveats here. First, this is only one study. Second, this is theoretical biology: it's not possible to measure the calories fat gives out directly, so this was established using indirect measurements and a theoretical model.

Third, the data for this study was taken from a 1950 study that starved 36 military men over a period of six months. I wasn't able to get a copy of the original, but found an informal report from a now-dead website called Mad Science Museum (yup, my hobbies lead me down weird rabbit holes). The report is quite a "fun" read (the experiment would never by approved by any modern ethics board):

The stress proved too much for one of the men, twenty-four-year-old Franklin Watkins. He began having vivid, disturbing dreams of cannibalism in which he was eating the flesh of an old man. On trips into town (before the buddy system had been implemented),he cheated extravagantly, downing milk shakes and sundaes. Finally Keys confronted him, and Watkins broke down sobbing. Then he grew angry and threatened to kill Keys and take his own life.

But the take-away for our purposes was that:

He designed the meals to be carbohydrate rich and protein poor, simulating what people in Europe might be eating, with an emphasis on potatoes, cabbage, macaroni, and whole wheat bread (all in meager amounts). Despite the reduction in food, Keys insisted the men maintain their active lifestyle, including the 22-miles of walking each week.

From what I know this is the abolute worse you can do to preserve muscle mass. An effective muscle-preserving mass diet should include plenty of proteins, as well as resistance training (to actually use the muscles — walking is no substitute).

This is not mentionned in the paper. On the whole, I'm not impressed with this paper at all, and it's kinda sad when a neophyte like myself can poke huge holes in a paper that seems to hold such sway — even the Stronger by Science guys quote it. I must say, though, that for a result that seems this important, it has preciously little citations (only 21 according to Google Scholar).

The starvation experiment also had an extremely long duration (6 months). I didn't dig in the math of the 2005 paper, but I wouldn't be surprised that this ends up screwing up the results as well, as the body will necessarily go into some kind of profundly weird dysfunction given such extreme conditions.

This study was popularized in the fitness community by one Lyle McDonald, to the point that the 31 kcal/lb figure is often quoted as "Lyle McDonald's rule of 31". However, Lyle McDonald himself now says that dieters can blow right past the 31 figure if they eat enough protein and perform resistance training. He also advocates the effectiveness of a short-duration (two weeks or less) rapid fat loss diet where the caloric intake is sometimes less than 1000 kcal per day.

In general, the reasonable consensus seems that the 22 kcal/lb figure can be taken as a safe lower bound for fat loss and that it is likely possible to exceed it without eating into muscles much for trained individuals.

In my case, given my starting weight of 90 kg and assuming 18% body fat, the safe 22 kcal/lb figure would have come down to a 785 kcal daily deficit, which is close to the 700 kcal deficit I got from the 0.7% body weight heuristic.

The idea that you can only mobilize so much calories from fat mass has one important consequence. Assuming you want to lose fat at the maximum "safe" rate, you will have to reduce your caloric deficit as your fat mass decreases!

But, on the other hand, there is the phenomenon of metabolic adaptation wherein your body becomes more efficient and is able to squeeze more calories from nutrients. So keeping a fixed deficit might not be a bad idea. As we'll see later, some step can be taken to combat metabolic adaptation.

Getting Serious & The Diet

But back to my cut. When we left off, I had completed 10 weeks of cut, including two refeed weeks, and lost 3.6 kg.

It's funny how memory goes: I thought I recalled hardship from that period, which is why I decided to increase my deficit to 1000 kcal per day. But this is actually not borne out in the numbers: in fact, I lost one kilogram more than estimated!

It is true however that I wasn't going as fast as I could have: I was generally undershooting the 700 kcal deficit goal by about 50 kcal, and the first four weeks I had vastly undershot the goal. I was hoping that the cut would not take much more than 3 months, and 9 weeks in, this meant only one month left!

Hence, the next two weeks I ran a 1000 kcal deficit... and nothing bad happened.

However, the increased deficit made my diet start being difficult. My daily energy expenditure (as reported by the Fitbit smartwatch) will vary between 2400 and 3800 kcal daily. If I sit in a chair all day with no activity at all, it will tend towards 2400. A lot of walking around or other activity assorted with an intense 2 hours training session will push sometimes push it in the 3600-4000 range (interestingly, there is a lot more variation at the top than at the bottom).

With a 1000 kcal deficit, it means that in a low-activity day, I could end up with a 1400 kcal budget at worst. This was problematic because I didn't alter my diet all that much: I kept eating the same kind of food I did before, making sure to get about 2g of proteins per kg of lean mass. So when you eat a 1200 kcal burrito for lunch, that doesn't leave a whole lot more calories for the rest of the day.

Regarding that 2g of proteins per kg of lean mass? I frequently undershot that goal, but I think I got at least 140g per day, which is already plenty and is said to be enough to prevent muscle loss. That being said, with the 1000 kcal deficit, it was clearly becoming difficult to even hit that diminished protein goal.

My answer to that challenge was to do something I hadn't done yet: add cardio to my training regimen on non-workout days. This ensured that every day saw a minimum of physical activity, making it easier to hit my protein goals.

After these two weeks, my weight loss was now 5.32kg. I also decided to have another refeed week. Again, I'm kind of puzzled as to why I did this. Sure, a 1000 kcal deficit is not nothing, but going by the numbers, everything was going according to plan.

I followed this last refeed week by two final dieting week, one at a target 700 deficit (effectively: 635 average) and one at a target 1000 deficit (effectively: 715). But for that last one I had the good excuse that it was the final week before my thesis' deadline...

Final tally: 7.02 kg lost in 14 weeks, including 3 refeed weeks.

Metabolic Adaptation & Supplements

As promised, a short discussion of metabolic adaptation. Metabolic adaptation is difficult because it does not have any single cause. When in significant caloric deficit, your body will silmultaneously become more conservative in its energy use, lowering your base metabolic rate (BMR) but it will also become more efficient and effectively squeeze out more calories out of the food you consume (you could say it "waste less" calories instead).

What can you do about it? There are many things you can do to mitigate the potential for metabolic adaptation that are basically good recommendations at any time t: eat enough proteins, get enough sleep, don't be too stressed. But there are two more active interventions.

The first is to break your diet and run a "refeed".

The idea is akin to that of a "cheat day" or "cheat meal", but ran over a longer period of time. The term can apparently reference different things, but in the kind of refeed I have in mind here, you eat at maintenance — so very different from a cheat meal where you're allowed to gorge on whatever food you like.

Cheat days apparently not enough to offset metabolic adaptation, hence the recommendation of running a whole refeed week.

Another important benefit of refeed weeks is that they offer a psychological reprieve in a months-long dieting process.

I should add that — having implemented ample refeeds — I saw no metabolic adaptation effects. Would there have been some if I hadn't done the refeed? Hard to say. These problems are known to be much more prevalent when you get to lower body fat levels (towards or beyond the 10% boundary for men).

If you want know much, much more about metabolic adaptation, check out The Metabolic Adaptation Manual.

The second intervention you can run is through use of supplements. There aren't really any magical fat loss supplements, with maybe the exception of caffeine. I managed to get my hand on the Examine's fat loss supplement guide and implemented what they recommended: caffeine, synephedrine, white willow extract and yohimbine.

I wasn't always consistent with it — I ran out of white willow extract at some point and had to reorder. Same with Yohimbine for the end of the cut.

Honestly, I'm not sure these did much, and my inconsistencies certainly didn't impact the data. The only thing that did seem to have an effect was increasing my caffeine consumption (through coffee). Caffeine has also other benefits for weight training that are well supported (one, two).

Caveats: given that I'm a relatively tall man, I might have needed to increase the recommended doses (which are given irrespective of gender, size or weight). I also read later that Yohimbine works better fasted and before workout — which is not at all how I used it.

Results & Intepretation

Okay, so I already gave the results, and much of the interpretation above. But there are still things to highlight.

So I lost 7 kilos. The final average global error was a mere -161 g — meaning that I lost 161 g more than predicted based on the recorded caloric deficit. During the process, the average global error went as high as -1000 g. Such variance is not necessarily unexpected.

The big question: was it all fat? Quite clearly (and visibly) fat was a big part of it. As for whether muscle was part of the equation, that's hard to tell as I didn't measure body fat during the process, except twice at they gym (once during on May 3rd (second week) at 17.2% and once on August 17th (two weeks after the end of the cut) at 14%.

We can also use the spreadsheet data to compute the bodyfat reached when assuming that 100% of the weight loss was fat. Supposing a 18% starting bodyfat percentage, the end result is 11.15%. This seems too low (mostly visually), but an alternative explanation to muscle loss is that part of this is water weight due to diminished carbs intake. I don't buy that explanation though — I weighted 82.7 on 17th August, about the same than at the end of the cut, and by then any carbs restriction effect should have been gone.

Another explanation: I didn't set my initial body fat percentage high enough. Setting the initial percentage at 20% brings the final percentage to 13.32% — closer to the 14% impedance scale measurement.

Another potential confounder here is muscle gain. When I started the cut, I was still making steady strength gains in the program I outlined last time — it's not unconceivable to have made some small gains here even in a deficit.

More relevant, I started training my abs and glutes much more vigorously (I wasn't targetting them at all before) during the last three weeks of the cut. There, I clearly made some muscle gains — at least a little bit of which I expect to have been made during those weeks.

Two more signs we can look at for muscle loss is fatigue and what happened after the cut. Regarding fatigue, I didn't really progress my heavy pre-existing lifts (deadlift, bench press, squat, etc) during the cut — but that was expected. In reality, I had to deload the squat and the deadlift them fairly early in the cycle for technique reasons (a further section of this article will talk about that — it's not connected to the cut). The bench and overhead press plateaued, while the rest of the program progressed rather nicely. I didn't feel tired at all.

Regarding what happened after the cut, my weight did not shot back up. I actually only had one week of training after the cut, and then a full week of holiday with no weight training (but a lot of walking). But even after, my weight didn't suddenly shot up even when I started running a caloric excess. This means there probably wasn't some easily recoverable "muscle memory" — but also that the body didn't overcompensate by storing lots of fat after the end of the diet.

So yeah... it worked rather well, but the jury is still split on whether it worked optimally (no muscle lost) or not.

Lazy Edit: So, if we ignore the issue of gained muscles AND if we assume that the caloric trackins is accurate, there might be a way to know, which is to solve the equations WL = FL + ML, CD = FL * 7700 + ML * 1540, where WL, FL and ML are weight loss, fat loss and muscle loss, while CD is the caloric deficit. 7700 is the number of kcal in a kg of fat, while 1540 is the number of kcal in a kg of muscle (thanks Colton for finding that info).

I might try to run this upon my data, but I'm not optimistic, given that the assumptions are pretty damn huge, and that the global error for the 100% fat model is pretty low.


What would I do differently next time?

First off, be much more careful with my tracking and discipline (knowing full well the occasional slip-up will happen).

Next, I would try running a ~900 kcal deficit immediately. Mind you, I'm not 100% convinced that this is the right idea, but I think it's worth trying for a 4 week duration. The benefit is obviously a shorter diet. The risk is that this is too much and muscle loss results.

Given my current body fat percentage and weight, a 900 kcal deficit is equivalent to requiring 77 kcal per kilo of muscle (35 kcal per pound) — which many people believe is safe. What is especially annoying here is that it will be hard to tell fat loss apart from muscle loss. I could use the neck-waist measurement method and the gym's impedance scale to try and estalish trends here, however.

I would also keep the refeeds after each 4 week period.

I can imagine playing with the length of the cut too. Assuming my body fat percentage doesn't increase (tall order), doing 4 weeks of cutting with a 900 kcal deficit would bring me to 10.5% body fat (assuming 100% fat loss). Two weeks would bring me down to 12.3%.

I would maybe drop the supplements, except caffeine. I might want to try to use Yohimbine "properly", i.e. in a mostly fasted state and before a workout.

Regarding caffeine, I would cycle off of it from for a few weeks before the cut, and then progressively ramp it up during the cut: drink something like one daily cup of robusta the first week, two the second, etc.

I'm excited to see if the second time is the charm for the six-pack :D

The Exercise Corner

How'd I do on my program meanwhile?

Let's first tackle the big lifts. The deadlift progressed to 185 kg until June. The squat was deloaded almost immediately, and kept being deloaded for poor form. In both of these lifts, I'd come to realize I had progressively adopted terrible form. I was desegmenting the deadlift: shooting my ass up and then pulling instead of pushing through the ground. I was doing something similar on the squat: bending forward too much so that my ass could go up and finally contribute to the lift. Both of these put a lot of strain on my lower back who had dutifully adapted. Symptomatically, deloading these lifts did not made them seem much easier after a bit.

After a bit of diagnostic, the issues seemed to be that my glutes and my abs were too weak. So in went a new regimen of ab training as well as glute-specific exercises, most notably the hip thrust.

These have worked wonder to strengthen these muscles, but I haven't had time to reload the lifts yet. And since I'm switching program for now, it might be a while before I see my old records (which I'm invalidating anyway, because of poor form).

The bench mostly stagnated, never really going much over that 100kg barrier, I managed 3x102 at times and then had to deload. The overhead press also stagnated and I also deloaded it. The problem there is arching the back too much, though I'm not entirely sure how to get rid of that — I'm hoping stronger abs will help.

The rest of the program mostly saw progress as some movements were still fairly new when I started the cut. Nothing went down, anyhow. Triceps, in particular, saw a lot of progress.

What's Next?

The wise man said: after each cut must come a bulk. And so it is.

I'm taking it somewhat more seriously than I did before. It feels like the careful attention to my calories during the cut as motivated me to be more serious now. I'm particularly attentive at ingesting more than enough proteins, and making an effort at timing them right too (something I started doing during the cut already).

Since I started reliably eating more than maintenance calories (aiming for 100-300 more), I sometimes overshot my calories goal whenever my caloric expenditure is a bit lower or on rest days. I must be careful with that. I think I have taken too much fat at the start of the year before the start of my cut by overeating. At that time, quite puzzlingly, I aimed to eat a certain amount of calories per day instead of basing it on the fitbit data. No idea why I did it like that ¯\_(ツ)_/¯

I must also pay attention to my sleep more. I've tended to push my sleep time further and further away and only my determination to at least get up for lunch has saved me. But often, sleep time ends up suffering in the process. I've been damn busy towards and since the end of the cut (you'd think handing in the thesis would mean work was over, but then you have to tackle everything that fell by the wayside on the way there).

I'm also trying out the Intermediate Build program from Jeremy Ethier. Since I got quite a bit of value from his free full-body program and his content in general, it seemed a safe value proposition. I'm currently tackling the first week and so far I'm enjoying it. We'll see come result time!

Finally, if I find the time I'd like to analyse my cut data a bit more. In particular, I'd like to run a multivariable linear regression on my data in order to see if I can find a systematic bias in the caloric intake or expenditure estimations. I'm not too optimistic here either: a non-naive model would include things like the hours classified by Fitbit as workout every day as well as their caloric estimations, as well as a labelling of how certain I am of my caloric intake measurements (things that come in a box are easy to measure, restaurant food is a wild guess).

A Note on The Fitbit

It is known that the Fitbit is not the most accurate means of measuring energy expenditure. If you look at the figures, this literature review paints a rather appalling picture, with many studies finding that the Fitbit under- or overestimates calories spent in activity by as much as 60% (and even more, in a few cases).

Of course, the fact that studies have such varied results is a big red flag. Either the Fitbit is totally random, or some studies are crap. Well, in my experience (and as can be seen in my data), the Fitbit is far from random. On the other hand, I have seen my share of crap studies.

If anything, the Fitbit was consistent — the same activity (in my case workouts and walking) tends to yield similar caloric estimations. If it does happen to be biased in certain cases, it looks like I was fortunate enough to have the biases cancel each other out.

Returning to the literature review, things look better when you look at the details:

Findings suggested that Fitbit was more likely to markedly overestimate energy expenditure when worn on the wrist and when walking at normal adult walking speeds on flat surfaces. On the contrary, Fitbit was more likely to underestimate energy expenditure when worn on the torso, with a tendency to markedly underestimate energy expenditure during inclined ambulation, during activities with constrained or variable body motion throughout the activity, and during simulated household or sporting activities that involve stop-and-start ambulation. Findings from 1 study for measures of energy expenditure in free-living settings suggested that Fitbit and doubly labelled water may provide similar measures of total energy expenditure over a 2-week period. However, findings from a few studies in free-living settings suggested that Fitbit devices may provide notable underestimations of daily energy expenditure compared with a SenseWear accelerometer.

Given my relatively high caloric expenditure numbers, I'm not too worried about underestimations.

Also apparently SenseWear overestimates energy expenditure?

I kept saying "the Fitbit", but of course there is no such thing:

Most of the studies included in this review were published in the last 2 years, with studies primarily examining measurement accuracy for models of Fitbit activity trackers introduced prior to 2015.

I used the Charge 2 (2016) until June and the Charge 3 (2018) after. It's possible things improved since 2015. I saw no difference in measurements between the Charge 2 and the Charge 3 however.

So what? Well, obviously take what the Fitbit says (and what everybody says about the Fitbit) with a pinch of salt. The best thing is to do like I did: collect your own data and come to your own conclusions.

Personally, I find the Fitbit super helpful to set a proper caloric intake.

What's the alternative? Using a macro calculator like this one. The caloric recommendation really depends on what goal (I chose "muscle gain") and "activity level" you set. With "lightly active" (true of me outside of workouts — and I did say I wanted to gain muscle, they don't build themselves), I get a 3185 kcal recommendation. Using my working hypothesis that the Fitbit is accurate, the average daily expenditure during my cut was 3190, and building muscle means I have to add 200-300 kcal on top of that. Oops.

On the other hand, choosing "very active" puts me at 4000 daily kcal. That's likely too much — I only hit 3700 kcal expenditure on workout days, and then not each time. So you have to choose "moderately active" to get a more reasonable 3591 kcal recommendation. How could I know this?

A promising avenue is the aformentioned Jeremy Ethier's program, which includes a nutrition spreadsheet. The initial estimate is wildly off (saying I should eat 3113 kcal for muscle gain) but it automatically uses your recorded weight and caloric intake to adjust the TDEE (total daily energy expenditure). I might try to use that and see if it gives good results.

And with these final comments, I'm done. This turned out much longer than I expected. Hopefully, someone will be able to get some useful information out of this!