[DannyBee]

and just to satisfy any curiosity, since i know 1500 calories of exercise a day sounds like a lot:

I was not fitbiting or any crazy mechanism to count getting up from the couch as exercise, or counting anything other than the actual exercise calories as "calories burned". Thus, it should have been a significant underestimate due to resting calories burned, etc.

In any case, it was running 5 miles a day, as reported by GPS, at 8:15 pace (which is roughly 120-150 calories per mile, depending who you believe, but i took the low side), and then running 1000 calories on an elliptical (which is about 1-1.5 hours at high resistance and fast pace, depending on pace).

The elliptical i used was one which slightly underestimates calorie count (by about 3%, at least according to two peer-reviewed studies), unlike most, which overestimate (by about 15-20%).

[dragonwriter]

While input-side variability may be a factor, if you are doing time-and-exercise-description without closer monitoring to determine your calories out, output-side variability is also a likely factor here.

Also, doing that much exercise, you can maintain a calorie deficit (a real deficit of calories absorbed vs. burned) and not lose weight by gaining muscle as fast as you are losing fat (muscle has a lower stored-energy content than fat); you'd need more than just stepping on a scale to distinguish that.