Improved Strategies for Betting on HI-LO

HI-LO, a routine, widely exploited by the gaming machine industry, was recently analysed using decision tree methodology. A disadvantage now recognised with the latter approach is that the calculations relating to whether or not players should quit or continue playing HI-LO, make no allowance for possible gains from future play. To overcome this drawback, the game has been reformulated for stochastic dynamic programming analysis. Key details are presented for the new methodology (based on backward recursion) along with corresponding results for an existing test application.