Applied Nate Silver – 5.2 Diet

Normally I eschew “fad” diets, even denounce, them but the 5.2 (feast/fast) diet appealed to me. Like all the other fad diets there is no scientific evidence to support it, but the idea is interesting. The notion is that continual calorie reduction forces the body to go into a starvation mode where the body cuts its calorie burning because it thinks it is starved. So alternating feasting and fasting fools the body – since it gets “normal” (and that is the rub) food intake on most days it avoids the starvation mode so that on fast days the calorie burn stays high at the same time as calorie intake is severely reduced. The other notion is that the “feast” days will not expand above “normal” so much that they completely offset the fast days and therefore the overall calorie intake will be below the calorie requirements and therefore you’ll run an overall calorie deficit,.

But does this work? Especially after having already achieved a substantial weight loss (about 70lbs) rather than early in a diet plan where rapid losses are “easy”.

Even though I’m skeptical about fad diets this idea might work for me because it’s easier for me to fast on a few days, then eat “normally” on most days than to have reduced consumption every day. As I’m nearing my target weight loss and thus beginning to focus on maintaining my new lower weight (the hardest part where few people succeed long-term) I’m actually interested in the 5.2 diet as a practical maintenance (rather than weight loss) plan. If I can just schedule 2 fasting days a week and then not worry about eating on the other 5 days that would be an easier plan than having to constantly watch consumption and then diet with greater calorie restriction if I happen to get an uptick.

So let’s apply a little Nate Silver analysis and see if my data proves anything.

So I’ve tried this for about three weeks and let’s see what happened. First let’s look at my weight history over a 20 day period where I fasted 5 days.


The first obvious thing is that my daily variation is large (i.e. there is lots of scatter around the trendline which has poor r^2 and thus not a particularly good predictor). The peak weights show a pronounced decrease but the lower weights do not – I wonder what that means (perhaps I’m really near the average of the lower weights as a “steady state” and just bouncing around).

But the real issue is whether there is any correlation between food intake, esp. on fast days, and weight loss for that day. I already know, from previous analysis, that my body is significantly dehydrated (on low days) so eating much of anything (high or low calorie values) drives up my weight, most likely just due to gain in “water weight”). Another way of looking at this is that my daily variation is a lot greater than any Δ in food intake calories ÷ 3500 cal/lb factor.

From start to end of the 20 day period I show a 4.1 loss, but keying off the trendline instead it appears to be more like 1.95lbs loss which fits closer to my other day (weekly weigh-ins). IOW, I’m running about 0.7lbs/week loss compared to about 2.3lbs/week loss for my first 24 weeks where I was maintaining an every day (with a few misses) 1200 average cal/day deficit (should produce 2+ lbs/wk loss, which is what happened). Now that my weight is much lower my daily calorie consumption would have to be about 400 cals/day less than when I first start (at my high weight), meaning either less eating and/or more exercise. But, in fact, May was a “bad” month for me: the lowest per day calories burned in exercise since I started my plan and a bit higher calories consumed, so it’s not surprising that I’m now doing 1.13lbs/week (over a longer timeframe).

But both my calorie consumption restriction (average per day) and my exercise (average per day) are greater than I can sustain long-term so it’s good that at least what I’ve been doing for past 5 weeks is still a weight losing plan since during maintenance it’s very likely I can’t sustain either my elevated exercise or my reduced eating. So perhaps this data suggests the 5.2 plan might work for maintenance (a lifetime of facing two fast days/week isn’t very excited, though, esp. given my “feast” days are still below my “normal” eating, but “normal” (in the sense of past 10 years) has to be reduced no matter what plan I adopt so my “new” “normal” has been be restraint and hunger every single day (even with the fast days). So the most the fast days accomplish is to make up for true “feast” days (i.e. the days where I break from a somewhat restrained consumption, i.e. my “new normal”).

So this might work, if I can really keep it up when I no longer have the incentive of actually losing weight, only just maintaining an appropriate level (for me, right on the borderline of “overweight” by the silly government definitions, but probably the one I must achieve for health goals).

Given all the scatter the next question is whether there is correlation between daily calorie intake and daily weight change (lots of up- and down-ticks during this period), so let’s look at that:


Again the r^2 here is very low and thus it’s not clear if there is really any correlation:

Hey Nate, a little help here, please

This graph is my daily Δ versus calorie consumption (as recorded with the BodyMedia app, fairly honest/accurate reporting) The cluster of fast results (all significant downticks) looks “significant” but the “feast” days are less clearly correlated with change in weight. But perhaps the trendline is suggestive – where it crosses the X-axis (no change in weight) about 1900 cals/day. According the various calorie calculators I require 2010 cals/day (no exercise) and 2450 cals/day (at my sustainable level of exercise) and 2740 cals/day (at my current level of exercise). So this doesn’t quite compute. But I know that my recorded calorie consumption will be almost always low (I simply don’t record every nibble), but I doubt I’m missing by only recording 2/3rds of my consumption (more likely closer to 90%). So my weight loss should be a bit higher than it actually has been.

Now that may just be lack of data to get statistically valid conclusion or it may be the effect that after the severe loss I’ve already had it’s difficult for my body to give up more, esp. as I am also now focusing on strength training and attempting to build muscle back up. Looking at my shape I still have some noticeable pudgy spots, but OTOH a lot of the body looks like starving prison camp inmate. Perhaps the pudgy spots are just left-over skin when my body diameter was 8″ larger and eliminating excess skin is a very slow (if ever, but I won’t do surgery just to look better cosmetically).

Now, ideally I could continue on this plan for a few more weeks and get at least twice as much data for better statistical significance, but alas, now I plan another interruption in this plan, another vacation (my April vacation led to 22lb gain in two week, which I also lost in additional 2 weeks). I suspect this trip (I’ll try to do a little better) will be a similar pattern, probably 15lb gain (camping is not a good opportunity to do careful calorie intake control plus I won’t have scale for monitoring plus fasting isn’t practical when I’ll need energy for outdoor activities). So that means another month will go by with no undistorted data, so it will be the end of July (hopefully July can be a focused month) before I can repeat this exercise.

Now while I’m at it let’s look at a few more bits of data from this dataset:

  1. Counting the fast days my average daily consumption was 1763 cals/day. That’s just a bit above my average during my more rapid loss during the first 24 weeks. This amount of calorie reduction, thus, should have produced more weight loss, but my exercise levels were also down by about 500 cals/day, which combined with lower requirements (at lower weight) probably explains the 0.7lbs/week vs 2.3lbs/week loss, also considering that rapid weight loss was easier when I was obese
  2. We had several large dinners and I indulged so my five highest days average was 2512 cals/day which is way more than even my highest days during my aggressive weight loss period. So “feasts” hurt (and require the fasts to compensate). During my aggressive loss I was targeting 1680 cals/day (and mostly hit that).
  3. My average for all “feast” days was 2133 so assuming I could stick with 2 days/week of fasting, I could probably boost my non-fast daily average to around 2474 (nominally what the calculator says I can have with even more limited exercise than I achieved during this period). Given some likely under-reporting of consumption this means I can probably achieve maintenance with modest daily calorie reductions (probably what nutrition scolds would consider “healthy eating”) + good (but not extreme) exercise + two fasting days (would be denounced by nutrition scolds).
  4. So, this might work. Show some restraint on “normal” days (compared to my old eating patterns), allow very rare pig-outs, stick with fairly aggressive exercise (as long as I can, good for BP anyway), and do two fasts a week (oh joy, they’re not much fun, but easier than reducing my 500 calories every day)

So, compared to my previous somewhat obese state while I was working (and less able to focus on weight) I need to continue:

  1. significant reduction (almost none) of my previous alcohol intake (probably used to average 400 cals/day, easy to cut that out)
  2. complete elimination of snack “junk” food (still can hit fast food joints, rarely)
  3. double the amount of exercise I did for two years prior to my weight reduction plan
  4. two fasting days

If I don’t do the two fasting days I’d have to go below anything approximating “normal” eating on daily basis and that would be hard.

Plus, I have to keep up my obsession with data collection, including daily weigh-ins, where any sharp uptick is the signal to force the fast days + increased exercise level (not hard to throw in, as long as my joints hold up).

So I think I can see a maintenance plan that is realistic and achievable.


About dmill96

old fat (but now getting trim and fit) guy, who used to create software in Silicon Valley (almost before it was called that), who used to go backpacking and bicycling and cross-country skiing and now geodashes, drives AWD in Wyoming, takes pictures, and writes long blog posts and does xizquvjyk.
This entry was posted in comment, prediction and tagged , . Bookmark the permalink.

2 Responses to Applied Nate Silver – 5.2 Diet

  1. Pingback: Applied Nate Silver – the noise is really LOUD! | dailydouq

  2. dmill96 says:

    I haven’t been able to stick with this, esp. hitting the 600 calories/day (usually a bit more), but also I’ve found that the unrestrained eating on non-fast days makes more difference. Of course the signal-to-noise problem of trying to measure real changes given the high daily variability makes this hard to determine. Perhaps the 5:2 makes more sense when you’re first started and not so much sense after 80lbs loss (and now at target).

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s