The chart at the top of this page is a graph of my weight loss program starting on June 1, 2012 up to my most recent weight entry. I try to weigh myself every morning, but if I go out of town, or just plain forget, I’ll interpolate the missing values after I get another reading.
The thicker, blue “weight” line is the raw scale reading. The thinner, red curve is a calculated “trend” weight. This is a technique I borrowed from The Hacker’s Diet. The trend weight is calculated something like this:
Tn = Tn-1 + ( ( Wn - Tn-1 ) × 25% )
In other words, take 25% of the difference between the current scale weight and yesterday’s trend weight, and add that to yesterday’s trend weight. This helps to smooth out the occasional bumps that can occur (sometimes just due to how much water your body is holding onto) by letting the previous readings contribute. As long as your scale reading is below your trend reading, you’re really still losing weight, even if it’s higher than what the scale said yesterday. This is similar to the statistical techniques that many experts use to analyze trends in stock prices (though simplified).