Abstract: Deal or No Deal is a television game show that involves one contestant, one banker and 26 briefcases containing different dollar values ranging between one cent and one million dollars. The allocation of dollar amounts inside the briefcases is unknown prior to the game to both the contestant and the banker. The contestant selects one briefcase to start the game that remains closed until the end. The game is played to a maximum of nine rounds, with a certain number of briefcases opened each round, revealing the dollar amounts. After every round the banker will submit a dollar offer to the contestant in an attempt to buy the contestant’s briefcase.
In this paper, the question of interest is how exactly the banker determines the offer for each of the nine rounds that will be given to the contestant. This paper will focus on the formulation of the banker’s offers. The data set collected by playing the online National Broadcasting Corporation (NBC) version and by watching the television (NBC) version of the games how are both analyzed to develop and compare several candidate models for the banker’s offers. These models will then be tested on new data points to determine how well the banker’s offer can be predicted for the online and television versions of the game show Deal or No Deal.