An NFL Survivor Pool is a popular contest centered on NFL game predictions that takes place during an active NFL regular season. Each week of the regular season, a contestant in a pool chooses 1 team. If that team loses, that contestant is out of the pool for good. If that team wins, the contestant remains in the pool and proceeds to pick a team during the next week of the season. Contestants can only choose a team once per season; if the team chosen by a contestant during a given week wins, the contestant cannot choose that team for the remainder of the season.
Pools can contain any number of contestants. The goal of the contest is to survive the longest in the pool, i.e. be the last contestant standing. Typically, contestants will pay to join the pool and the winner of the contest is awarded the money collected by all contestants.
Contestants must strategize to survive the longest in their pool. Some contestants may simply choose the strongest team each week, which is commonly known as the “greedy” method. Though this may result in a contestant surviving early weeks in a season, a caveat to this approach is that a contestant may be left to choose weaker teams later in the season. Alternatively, contestants may have a more conservative approach and save stronger teams for late in a season. This is imprudent as strong teams do a contestant no good in later weeks of season, if the contestant is eliminated early due to choosing weak teams. Contestants must be diligent in strategizing to win an NFL Survivor Pool. The goal of this project was to explore the concept of NFL Survival Pools and create a tool facilitating the optimal choices contestants should make when participating in an NFL Survival Pool.
Problem Statement
Team ESPN (Audrey, Lauren, Adam and Dominique) were challenged to create a tool that provides guidance to the problem:
In any given week of the season, what team should an NFL Survival Pool contestant select to optimize their chances of surviving the longest in the pool? How would this strategy compare to common heuristics, such as the “greedy” method of survival?
ESPN currently does not host a service to specifically address these questions. ESPN desires to remain prevalent in the sports industry, and this tool would add to their portfolio of sports analytic services and promote their popularity amongst sports fans.
The Sponsor anticipated to use the tool as the structure for a service that ESPN could provide to NFL Survivor Pool contestants.
Requirements
The primary requirements for this project were:
The team will create an integer optimization model with the goal to maximize the serial probability of winning by surviving longest in an NFL Survival Pool.
The model shall account for uncertainty regarding the relative strength of each team, both currently and in the future.
The model will provide a suggested team for any week of the season, given the teams a contestant has already selected.
The team will work to demonstrate the superiority of the model compared to common heuristics, such as the “greedy” method.
An NFL Survivor Pool is a popular contest centered on NFL game predictions that takes place during an active NFL regular season. Each week of the regular season, a contestant in a pool chooses 1 team. If that team loses, that contestant is out of the pool for good. If that team wins, the contestant remains in the pool and proceeds to pick a team during the next week of the season. Contestants can only choose a team once per season; if the team chosen by a contestant during a given week wins, the contestant cannot choose that team for the remainder of the season.