Predictive Analytics for Real-time Auction Bidding Support: a Case on Fantasy Football

Abstract

This work reports about an end-to-end business analytics experiment, applying predictive and prescriptive analytics to real-time bidding support for fantasy football draft auctions. Forecast methods are used to quantify the expected return of each investment alternative, while subgradient optimization is used to provide adaptive online recommendations on the allocation of scarce budget resources. A distributed front-end implementation of the prescriptive modules and the rankings of simulated leagues testify the viability of this architecture for actual support.