Project Selection: Portfolio Analysis using Copilot in Excel with Python

Brian, it’s fascinating to see portfolio management evolving with Copilot in Excel. But the real question isn’t whether we can automate models that are over 50 years old—it's whether we should even be trying.

I developed the first version of this model in 1995 and sold it to Microsoft in 2007. The portfolio selection and efficient frontier screens still bring back memories. Over the years, prioritization has been widely used to rank strategies and programs. But optimization, while mathematically elegant, struggles beyond narrow domains, which is why it hasn’t seen broad adoption.

The difference between old tools and what we need now isn’t just better and easier automation—it’s the ability to have a conversation with the model, to explore a much larger domain. Traditional models force binary decisions: select or reject. But in reality, most projects sit in the gray zone where trade-offs matter. What we need are tools that enable dynamic conversations, where teams can understand risks, deal with uncertainty, and collectively reach decisions about where to invest—not just rely on a mathematical model to dictate outcomes.

With Copilot, we have the chance to rethink this approach. It’s not about making faster decisions—it’s about making better, more informed ones. Should we continue automating old methods, or should we embrace tools that allow us to fully explore the complexity of modern decision-making?