In a previous post, I claimed numeric projections should be distributions. This is an example of something that can be built on top of such a projection system. It is an evaluation of a 12 team draft for a 5×5 fantasy baseball league.
Team 3 has a 30% chance of winning the league with this roster:
Evan Gattis Hou – C
Paul Goldschmidt Ari – 1B
Dustin Pedroia Bos – 2B
Kyle Seager Sea – 3B
Carlos Correa Hou – SS
George Springer Hou – OF
Giancarlo Stanton Mia – OF
José Bautista Tor – OF
Todd Frazier CWS – 1B,3B
Jake Lamb Ari – 3B
Carlos Martínez StL – SP
Cole Hamels Tex – SP
Jeurys Familia NYM – RP
Héctor Neris Phi – RP
J.A. Happ Tor – SP
Brandon Kintzler Min – RP
Chris Tillman Bal – SP
Jon Gray Col – SP
Distributions give more traction for calculating probabilities, all the way up the chain from player projections, to team-by-team category projections, and finally, probabilities for winning the league. I am working on a short essay on decisions to expand on these ideas, as well as an automated system for evaluating drafts.
If you want to be notified when this system is up, please submit this form:
*quick note on the points histograms, the labels on the x-axis should be divided by 2