overview

This project is a work of analytics, aiming for simplified estimation. And I'm talking about fantasy football here. The model takes only two input variables: past stats and the NFL draft. Coaching philosophies, training camp videos, scouting reports, trade rumors - none of these receive considerations. In this way, produced are unbiased estimated values not subject to caprice.

With that said, this site offers dynamic results depending on which metric you feel fits each player. League differences are also respected. Your trades are judged according to your own league's settings.

The trading model uses regression techniques in a sort of mixed model. One part employs the ordinary least squares of simple linear regression, coefficients of which undergo smoothing from a weighted least squares, multiple linear regression function. Training data is made of draft results and NFL statistics from years 2005-2020.

For a detailed explanation of the process, follow the links to continue reading; otherwise, be joyful about your business.

    >