METHODOLOGY
WHAT THIS IS
Every metric on ATROPOS is calculated using millions of full-contest simulations: each game modeled using market odds and every entry making a pick each week until eliminated.
REALISTIC MODELING
Simulated seasons unfold like normal NFL seasons: injuries hit, team strength shifts, and lines move. The field reacts each week and picks accordingly.
FIELD MODEL
Simulated opponents are modeled on participant behavior: saving teams for their strongest spots, weighing win probability, and prioritizing survival each week.
TIER DIFFERENTIATION
Each week, teams are placed into five tiers using win probability, projected pick rate, future value (derived from our proprietary FATE score), and simulation results. Tier 1 represents the best overall options, not necessarily the biggest favorites, balancing immediate survival with value later.
ANANKE™ SIMULATION ENGINE
ANANKE powers our models. Its simulations are calibrated against historical NFL line movement and run under each contest’s rules. Millions of full-season simulations identify the strongest routes, which are then tested on separate simulation sets. Published EV is measured out of sample: no route is scored on the same results that selected it.
METRICS
WIN
A team’s win probability for each game, derived from real-time market odds. Unposted games are projected from the matchup’s latest available line, adjusted for subsequent market movement.
PICK
The projected share of entries selecting each team that week.
AVAIL
The percentage of entries that can still use the team.
WINSHARES
A team’s share of the week’s combined win probability in excess of 50% (0% on holiday weeks). Used to measure option depth and identify each team’s strongest weeks.
EV
A team’s expected value for the current week, based only on win probability and projected pick rate. Future value is captured separately by FATE.
FATE
Field-Adjusted Team Equity. A team’s future value based on the strength of its remaining spots, calculated separately for Circa and Splash formats.
EDGE
Our proprietary measure of a pick’s value relative to the median pick, accounting for future decisions.
TIER
Tiers are ordered by the best overall options, not necessarily the biggest favorites, balancing immediate survival with value later.