The causes for the failure of complex IT projects are well known in the computer industry. Assessment of project risks typically include metrics gauging team performance, stakeholder commitments, schedule adherence, among others. The process of obtaining an aggregated delivery risk score for the project involves assigning a score to each of the individual dimensions and then aggregating the results either linearly or through a more complex aggregation function. The outcome is typically an ordinal grade such as "A" through "D" where "A" denotes a high performing project while "D" denotes a poorly performing project. Thus, project performance studies typically look "backwards" at the historical trend of root causes, as they evaluate "what went wrong." Furthermore, this measurement is quickly outdated and the process of assigning scores to project performance measurement can only provide a point-in-time- measurement. Given the overhead involved in repeating the assessment, the process of evaluation is repeated once a quarter or less frequently We have developed a predictive algorithm which gives the risk manager a score indicating the likelihood of project success or failure before the project begins. This opportunity allows us to quantify the "odds" of the next troubled project, thus supporting managers in effective risk mitigation.