Friday, February 22, 2013

Weight Graph Analysis Using TA

The attached chart is a great example of the insight financial technical analysis can bring to traditional weight charts. This particular chart plots my daily weight values in blue over the last ~11 months, and after a quick glance it becomes clear that the day to day values can be quite volatile. Two of the most common security charting techniques, the moving average and Bollinger bands, can be quite useful not only in making sense of this variability, but by also helping me keep these fluctuations in perspective. As anyone who has tracked their weight can tell you, there are times when you are doing everything right in terms of exercise and net calorie reduction, yet the pounds don't seem to come off as one might expect. This is where Bollinger bands can be useful.

Bollinger bands represent 2 standard deviations added to and subtracted from the mean, which happens to be my weekly average weight in this case (shown in orange). In a normal distribution, around 95% of the sample population is expected to fall within these values, which are represented on my chart as the grey lines above and below the daily and weekly average values. While weight values aren't technically normally distributed (or stock prices either for that matter!), the use of these bands are still quite useful in the sense that most values should still be expected to fall within this range. As my day to day weight change values begin to increase or decrease in size, the bands reflect this by widening (for an increase in volatility) and contracting (for a decrease in volatility). So when I am chugging along with my weight loss plan and the pounds are coming off quickly, the odds of a relatively dramatic subsequent increase in weight are actually increasing. This interpretation becomes clear when you look towards the beginning of my chart, when my weight was swinging widely between 185 and 195 lbs. After an initial spike up, the pounds came off just as quickly and almost to the same extent that they were added on. The tendency for weight values to behave like this is quite common, with the moving average serving as the center point of this oscillation, and I eluded to earlier, the use of moving averages on weight charts is one of the best ways to keep daily values in perspective.

Moving averages smooth data so that the short term volatility is eliminated, at the cost of some responsiveness. A moving average can be thought of as a "typical" value for the period in question, one week in my case. What is interesting about the moving average is the relationship of the daily values to the moving average itself, as the daily values tend to cluster either below or above the average and tend to stay there for similar intervals. This ebb and flow is the natural process by which we lose and gain weight, which can be comforting to someone attempting to shed some pounds. The red bar chart below my weight graph is the perfect visualization of this phenomenon. This chart, "7-day average daily weight loss", plots the cyclical nature of this process while also bringing to light the portion of time I was actually gaining weight, which is much higher than one might initially expect (during a phase were my net weight loss was pretty significant). Obviously the goal is to drop your overall weight, but the process by which this is done is fundamentally a give and take relationship: you lose some, you gain back a little less than you lost. The point is not to be discouraged by the counter-moves during the process.

No comments:

Post a Comment