The Sim

DraftKings Pricing. Updates throughout the day. Scroll down for further detail.

Sim v3 uses data from each individual player to build models that update their expectations after seeing each new game's data. These models are then adjusted for factors that wouldn't be predictable from past data (Line, PP, expected TOI, Vegas odds, etc.). Together, the output is very similar to v2, providing probabilities to exceed expectation thresholds and by how much. The difference this year is how we get there!

      - Doug


Key

Let me take you through the new features and how to use them:

  • n - games used to train the model for each player
  • salexp - Showing you the expectation based on salary for a DK player; Exp = Salary/1000 x 0.6
  • adj - Adjustment factor for game-situation. Basically, this tells how how much better than base performance a player is expected to perform in his given spot. This takes into account Vegas, Line, and PP especially.

Adjusted Stats - Adjusted versions of your old favorites are here using the adj factor in conjunction with the base model trained on each individual player.

  • Freq - Frequency of scoring above salary-expectation. This is a great Cash stat because it adjusts expectations based on salary, so even a low salary guy gets love if he is consistent. High frequency = Consistency
  • Mag - Magnitude of upside: when this player scores above his salary-expectation, how much does he score? In my opinion, this stat is better than a strict "Points" projection because of the bimodal distribution of player outcomes (i.e. players either score lots or none).
  • PICL - Player Index for Cash Lineups. Uses an increased weight of Freq. to more accurately predict the value of putting any player in a cash lineup.