While much research has examined project management methods and their ability to handle complexity and change, little such research exists at level of project portfolio management. This is somewhat surprising given that portfolio management must, by definition, cope with the cumulative changes and complexities of the projects contained within it. Complex adaptive systems theory (CAS) has provided useful insights into the management of unpredictable emergent system level properties in a number of disciplines. This exploratory study uses 30 expert interviews with complexity scholars and IS portfolio practitioners to identify portfolio management practices from emergent properties of real world CAS. The findings show that portfolio managers can learn from CAS how to manage and shape emergence. 15 complexity based practices are identified, along with examples how experienced practitioners achieve them. For example, just as a bee-keeper can predict swarming, portfolio agents can be trained to identify weak signals predicting major change.