Operationalizing the Kelly Method to Bet on an Innovation Project Portfolio

In this article an actual innovation portfolio is used to test the “Kelly Method” for allocating resources among a set of projects in a portfolio. The portfolio could include innovation projects, capital investment projects, or even mergers or acquisitions. The Kelly Method treats the investments as a type of gambling bet, in which the bets can be favorable, and optimizes the median (not mean) return on the portfolio of bets. The method is operationalized to examine an actual portfolio. Using existing corporate numbers, the relevant “ptab” parameters are estimated: Probability of success, Time until launch, Adversity ratio for losses, and Benefit ratio for wins. An optimization routine then identifies the mix of projects─including fractional projects when that is optimal─that gives the highest profitability for the entire portfolio. Part of the operationalizing is in adding constraints such as a bucket for “seedling projects” (i.e., low maturity, low probability), to keep the innovation pipeline full. The Kelly Method is compared to a “Rank ROI” Method, in which the Return on Investment is calculated from the estimated parameters, and the projects ranked by the Present Value of the expected ROI. Similarly, we examined a “Rank NPV” method. An important conclusion is that the Kelly Method gives the highest profitability of all the methods for this portfolio, higher than Rank ROI or Rank NPV. This first step in Operationalizing the Kelly Method for budgeting among projects in an innovation portfolio shows the importance of translating as much qualitative information as possible into numbers, which also helps in cross functional communication.