Using Mock Draft Data to Create a Player Availability Dashboard for the NFL Draft

The NFL Draft is not only important for teams, but also popular for fans, and its popularity incentivizes sports media entities to create “mock drafts” leading up to the draft that attempt to forecast which players will be selected by the different teams. This project took historical projections from a variety of public mock draft sources and attempted to quantify the historical accuracy of these mock drafts by approximating the error distributions around these projections. These historical error distributions were then combined using Bayesian inference into a final probability distribution that represents the likelihood that a player will be selected at any particular point in the draft. Where there seems to be a consensus around players from multiple sources, the resulting probability distribution is tighter and more certain around these picks. Where there is much disagreement on the projection of a particular player, these distributions are wider and less certain about when a player will be selected. These probabilities are found to be beneficial in that they are not only more accurate than relying on a single source of mock drafts, but also provide a more complete picture of the range of outcomes of a player than a single spot projection. Finally, these probabilities for when a player might be selected were organized in a Tableau workbook along with other draft-specific data that could serve as a dashboard for a team during the NFL draft to build their draft strategy.