SweatStack Power-Duration Model Fitting

Your mean-max curve is your best recorded power at each duration. A power-duration model fits a smooth curve through it, but the data carries two kinds of error:

Measurement noise. Small (typically within 1-2%) random errors from your power meter, scattering points above and below the truth.
Submaximal efforts. This error is one-sided: a point can only sit at or below your true ability, never above, and this kind of error dwarfs the noise.

The standard method to fit a model (least squares) sits in the middle of the cloud. That's right for symmetrical measurement noise, but it can't tell a submaximal point from a low reading, so the model fit gets dragged below numbers you've already hit. That underprediction is plainly wrong: it says you can't do something you already have.

SweatStack fits the upper edge of your mean-max curve instead, so the model sits on your proven bests and never tells you you're weaker than you've already shown. Keep in mind that like any mean-max model, it assumes you went all-out somewhere.

Loading sample data…

SweatStack Power-Duration Model Fitting

Power vs. duration

Underprediction