Applied Nate Silver – stuck in the ’90s

No, not the 1990s – the 190′s of my weight trajectory. It feels like I’m stuck in this particular decade and hardly changing. Through four previous “decades” (i.e. 23x->22x) it has taken me four weeks to pass through, but the 19x decade feels stuck, three weeks of 19x weights and not even halfway through yet. Am I hitting some sort of plateau? Are I running out of steam in my reduction plan?

This may also be an illusion because I’ve been here before. I don’t have precise memory and all this detail, but 25+ years ago when I was training for triathlon I lost weight too, starting somewhere in the 220′s (far lower than this time) and then “stalling” in the 190′s. So perhaps since that happened before I just think it’s happening now, again. (Is this example of ‘confirmation bias’?) That past time, poorly remembered, may also not be at all relevant since then my body mass was entirely different (a lot more muscle, less flab) and my training (more intense, more focused on upper body) was different and, of course, I was a lot younger. Or perhaps the 190′s do represent a bracket that is hard to break through? And am I just alarmed for nothing and this is a minor statistical blip? Maybe now my excess weight is mostly skin that is slower to shrink that adipose? That’s the whole point of looking at this from the Nate Silver POV, is this a signal or noise?

So let’s look at the data, and as usual, my plea:

need some help here, Nate!

(maybe he’ll finally respond):

weekly23-190s

This is just my average (of multiple weigh-ins) daily value since first breaking into the 190′s. At first glance it doesn’t look like it’s stalled. The points clearly have a downward trend as indicated by the regression line’s slope of 1.68 which contrasts with 2.39 (weekly, since the beginning of weight loss) or 2.25 (weekly, multiple measurements on weigh-in day) or 2.39 (since start of daily weigh-ins) or even 2.16 (composite measurements, weekly spreads and daily spreads) for half the time of this plan. Now, of course, longer-term trends don’t translate into short-term trends, but I’ve not (fundamentally) changed my weight loss activities (diet or exercise), at least consciously. And that is the point now, have I actually changed my plan without really meaning to? Or is my plan just less effective about 60lbs lost?

Now I’ve been “alarmist” when I’ve seen blips in my data before. When I first had a small unexpected uptick I discovered the variability was merely due to scale and thus began to take multiple measurements for my weekly value. Then I had a large unexpected uptick on my once-a-week weigh-in and started taking daily measurements, to deal with anomalous weights. But that also started a bit of panic and rededication to extreme dieting which may have accelerated my rate of decrease. In short, all these stats (and possibly misleading interpretations of them) have triggered some change in behavior.

So, for instance, when I had disappointing results on the daily’s last week and thus anticipated my weekly loss would be well below plan I re-dedicated myself to really pushing hard, avoiding any of the eating lapses I had last week. So yesterday, for instance, I was below my “quota” (as defined in the BodyMedia plan) on eating and well above on exercise, although BodyMedia only shows a slightly above-average calorie deficit. And I watched any circumstances that might affect “water weight”. So what was the result?

Here’s my rolling 8 day graph:

weekly23-190s1

The red dots are the average of the daily samples and the blue dots are the raw data. It is the values of the red dots that appear in the first graph. So for today, as measured by the red dots, there appears to be a significant reduction. BUT, it’s probably just an illusion. Look at the spread of blue dots for today (rightmost data) and yesterday. It’s just the scale variability that changed (stdev = 0.74 yesterday vs 0.33 today) and as measured by the lowest values today was actually a slight gain. Certainly it’s hard to justify the POV of today as a decrease, despite a 1548 calorie deficit yesterday. And, in fact, today’s point (and removing the 8 day earlier point) switch the trendline slope from positive (gains) to negative (now small loss, 1.1, even less than the 190′s range (1.7). IOW, the last seven days have been bad. But I knew that, last week was an exception in diet.

But is all this just a short-term blip that hopefully I can reverse this week assuming I manage to avoid any unusual eating. Let’s look at some more data:

weekly23-190s2

This is calories burned, as measured by the BodyFit sensor and BodyMedia app. The jiggly line is 7-day moving average. The trendline, despite the low r^2, is probably real. But all this, I think, is an artifact. I have been adding my weight loss to the BodyMedia app. The way most formula calculate calories burned is based on weight. Even though BodyMedia doesn’t provide any transparency on their calculations (a bad thing, they should, but generally BodyMedia is more hucksterism than good science) I’m going to guess that the drop in calories burned is just due to drop in weight being used in the calculations. Now that effect is real and is part of the whole plateau thing (as you get more fit, the same amount of exercise burns less calories), but it doesn’t seem to explain it.

So here’s some more, calories consumed:

weekly23-190s3

No clear trend here. The r^2 is way too low to even believe the trendline. But what is very noticeable is the big change in variability. In the earlier part of the graph I’m fairly consistent, then I went through a panic and briefly substantially reduced my consumption. Then I went the other way and had some extravagant days. In short, my control is erratic, so food consumption has not been helpful and probably does explain the 190s delay.

So let’s just look at that, even though the numbers aren’t very trustworthy. Why? Well, the BodyMedia thing is fairly useless for tracking. Its measurement of calories burned is very poorly correlated with exercise, so that’s wrong. But calories consumed is my self-reporting and often based on guesses and almost certain to be more likely to be under-reported than over. In short, the error the calorie deficit is large, but let’s look at it anyway.

weekly23-190s4

I think this shows two things: 1) as shown by 7-day moving average, my calorie deficit has really decreased and obviously I won’t loss as much weight when I don’t have as big a calorie deficit (either exercise or diet), and, 2) it also shows the much greater volatility which is that I’ve drifted away from a more planned diet to a less controlled one (making meals, eating out). What I need, clearly, is to stick with as rigid as possible consumption and as low as possible, but possibly just more consistent alone is enough (maybe fat burning is erratic with erratic calorie gaps).

So the delay in the 190s is real but not entirely explained. What I need is two weeks of solid stick-to-plan effort, more exercise, less eating. The trouble is that’s not what I can fit into my life.

I’m now in the range where preparations for the vacation begin to take too much time and thus reduce my exercise. And then the vacation itself will be terrible for controlling consumption. So at best I may get a little drop this week and then everything falls apart. And April has several interruptions to just focusing on weight loss, so I suspect I will be stuck in the 190s for at least 5-8 more weeks. AND, I may be lucky to be stuck there (instead of going back up, which I do expect some temporary blips). So it could easily be all the way until June before I break out of this band.

And where do I need to be? Well, I can feel, as well as measure, that my flab and BMI is still too high. While it means reverting to my weight from 30+ years ago I need to be in low 180s. It’s amazing I can do 60lbs relatively quickly and now face months just to do another 15lbs, but that is a probably with weight loss – life interferes. I have other priorities and so I can’t just intensely focus on this for another 6-8 weeks, which should be enough to achieve my goals if I could focus.

So this sucks, to fall short of goals just before entering a time where sticking to plan is going to be very difficult (I won’t even have an measurements for 10 days) or impossible. And if by the time I can focus again I’m back up by 10lbs that is going to be really discouraging. So I need a “success” before entering that period of uncertainty.

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, musing, whine and tagged , , . Bookmark the permalink.

3 Responses to Applied Nate Silver – stuck in the ’90s

  1. dmill96 says:

    Being stuck is really driving me crazy. The day I wrote this post I managed a huge exercise and thus huge deficit. Weight next morning? No statistically significant change (up slightly but within one sigma). So today I’m trying for the best possible number – total fast and as much exercise as I can spare time for. Whether this barrier is psychological or physiological I think if somehow I can punch through it maybe I can get back on track. It’s hard to do total fasting just for a short-term goal, but this is important – I must get a lower value. Pi did it. No matter what mechanism is halting my losses days of zero intake has to do something. But can I actually do that?

  2. dmill96 says:

    And Thursday, after only consuming 300 calories all day on Wednesday, yet still doing over 1000 calories on treadmill, finally showed a drop. But the drop is way too large. The calorie deficit, at most, would justify 0.8lb drop, but it actually 2.6lbs, obviously not true (while in a good direction it’s just as bogus and therefore should be as upsetting as the anomalous upticks I’ve had. Upsetting because it simply means I don’t actually know what the real weight loss was and therefore whether I’m back on plan or not).

    Now I’ve had two strong and unjustified downticks. The circumstances are entirely different – one was on a day with relatively large calorie consumption, the other on record low consumption. So is there anything in common? Yes, almonds.

    Of course the nutrition freaks are into nuts as the new wonder food. Of course, it’s implicit in their thinking that anything white (and therefore salt) is evil. So undoubtedly almonds can only be good, unsalted, and of course, raw (except that raw will actually kill you, roasting is required for almonds, but raw is dogma among nutrition nazis, even though it’s often wrong, and really wrong in this case). But I believe the issue is, in fact, salt. Now given my BP is almost too low, I’m not worried about salt as a health issue, BUT, salt clearly affects water retention (why am I now into brining many meats except to increase juiciness). So in my case I think it may be salt, not as evil and going to kill me, or that it interferes with weight loss; no, salt is just noise obscuring the signal I’d like to measure more accurately.

    Now, since I’m starving I’m wondering if there is an experiment I can do. I do have some unsalted almonds, uhm, shall I eat those, at the same amount as the salted ones and see what happens? Or would eating anything, other than doing almost total fast again, bloat me back up. I think tomorrow my weight will be up, no matter what I eat today, so once again noise would obscure signal, so if I want to do the unsalted almond test I need to do it on a different day.

  3. Pingback: Back at Starbucks – 6 | dailydouq

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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