Dynamic programming and the evaluation of gaming designs

Since their invention before the beginning of the last century, slot machines have evolved to become the most profitable form of gambling activity for land-based casinos. Now the potential exists for an analogous development in the online gaming world – where the number of virtual ‘slots’ is effectively limitless. In this article we focus on the economic evaluation of slots – particularly that of a mainstay routine of many contemporary gaming designs, HI-LO. Decision tree analysis was used for a previous evaluation of HI-LO. In contrast the assessment that follows is based on a stochastic dynamic programming (DP) (backward recursion) analysis – the default methodology in the industry. Corresponding results for a real-life test application confirm the superiority of the DP approach.