It's great to be back for another season. Hopefully everyone here is happy with their drafts. If not, no worries, season are won by playing the waiver wire.
I'd like to remind everyone that relative prediction accuracy during the season looks something like:
80-85% for wide receivers and running backs
75% for quarterbacks and tight ends
60-70% for defenses
50-55% for kickers
Remember this when managing your roster, and good luck in week 1.
I'd like to remind everyone that relative prediction accuracy during the season looks something like:
80-85% for wide receivers and running backs
75% for quarterbacks and tight ends
60-70% for defenses
50-55% for kickers
Remember this when managing your roster, and good luck in week 1.
Introduction
In this fantasy football model, I take aggregated expert ranking data from Fantasypros.com and pass it into a statistical clustering algorithm called a Gaussian mixture model. The algorithm finds players who are ranked similarly and discovers natural tiers within the data.
The results are a series of simple charts that can help you decide your fantasy football line up each week. Match ups, injuries, strength of offense and defense are all accounted for. For more intuition behind the model, you can read this introduction.
Questions or comments? Tweet at me @borisachen.
How to use these charts
(1) Favor higher tier players.
(2) When choosing between players of the same tier, flip a coin or go with your gut. If experts are 50/50, there's no wrong choice.
(3) Fill your non-Flex positions first. Then use the Flex charts.
(4) Charts update every 3 hours. Check back Thursday, Sunday and Monday before game time.
(4) Charts update every 3 hours. Check back Thursday, Sunday and Monday before game time.
Other charts are accessible from the menu.
