The author of The Hacker’s Diet, John Walker, has generously given me permission to discuss his diet in detail using some quotes, with links back to his site. (Earlier E4T articles on The Hacker’s Diet)

Floats and sinkers

Next, consider the trend line along with the daily weights from the scale.

The trend line is drawn in as before, and each day’s weight is plotted as a diamond. I’ve drawn lines from each weight measurement to the trend line to show the relationship between daily weight and the trend that day. Remember, in an exponentially smoothed moving average the most recent day’s measurement has the greatest influence on the trend line. Recall also, that since the moving average looks back in time, it lags the actual trend. Consequently, when the trend is falling, most of the daily weights will be below the moving average trend line. Think of the trend as a fishing line in the water. Daily weights that fall below it are sinkers, pulling it down; the further the weight is below the trend line, the stronger it pulls the trend line down. When the trend is rising, most daily weights will be above the trend line: floats, tethered to the line, pulling it up.

(snip)

The relationship between the daily weight readings and the trend line is not only a powerful psychological tool in controlling weight, it also provides early warning about shifts in calorie intake and therefore changes in the trend of weight. While you’re losing weight, you’ll see a chart that looks like Dexter’s: a falling trend line with most daily weights below it.

(snip)

As you can see, the computed straight line trend accurately reflects the rate of change in the moving average. Fitting a straight line to the trend doesn’t eliminate the need to interpret the weight loss or gain and calorie excess or deficit in conjunction with the chart, taking into account the shape of the trend line. Consider this chart of Marvin’s diet, showing the moving average and two very different straight line trends that can be calculated from it.

The familiar moving average trend line is drawn as before. The dashed line is the best fit to all 90 days of Marvin’s diet. It’s accurate, after a fashion, since Marvin ended the diet below the weight where he started, and he lost weight for 50 days but gained only in the last 40. But, obviously, it misses the point. What really happened was a sequence of successful weight loss, stability for a brief period, then rapid weight gain. Short term straight line trends, in this case month by month, identify the turnaround and provide accurate estimates of weight change and calories for each month.

As long as the moving average is roughly straight, the weekly weight and daily calorie figures can be relied upon. If there’s a major change in trend during the month, you’re better off basing your actions on the most recent direction of the trend rather than longer term values that don’t account for a kink in the curve. . . .

Ahhh — At last we’re pretty much done talking about weighted moving averages!

He talks more about it (Amplitude, rate, Trying trend fitting) using some of his spreadsheet tools. But I’m anxious to move on to the next topic (and I barely understand this much.)d This summary covers the topic to the point that I got lost. And I’m hoping that if I revisit the topic later it might be more understandable to me.

Meanwhile I like what he says about the encouragement and warning we’ll get from the ‘Sinkers’ and ‘Floaters.’a And I appreciate the warning about trendlines that are too long.