Stochastic Roll-Up: Taming Multi-Project Risk | The Risky Planner | S2E22

Send us Fan Mail (https://www.buzzsprout.com/2451045/fa...) If your program sets contingency by summing project P80s, your confidence interval is lower than you approved. No AACE standard addresses this. Albert Brier, Roger Bradfield, and Rachel Fleming present the only framework that does, June 30 in Las Vegas. Every organization running a multi-project program is making the same mathematical error when they set their contingency budget. Not because their people cannot do the math. Because no formal standard exists at the program level. The AACE has six recommended practices for project-level risk estimation. It has zero for programs. Nate Habermeyer and Albert Brier are joined by Roger Bradfield, the show's first guest, to discuss RISK-4852, a contingency management framework for multi-project programs co-authored by Albert Brier, Roger Bradfield, and Rachel Fleming of MBP Consulting. The episode covers where the gap in program-level standards comes from, why adding probabilistic contingency estimates across projects produces the wrong number, and what the stochastic rollup method does differently. The paper presents at AACE International 2026 in Las Vegas, June 28 through 30. The problem starts with a simple question nobody can answer. When a client in a monthly program review asks for a P80 value across the entire portfolio, the answer should be straightforward. In practice, most program risk teams either sum their individual project contingencies, which is mathematically incorrect, or build a monolithic program schedule with tens of thousands of activities, which is practically unmanageable. Neither approach produces a defensible program-level confidence interval. The framework Albert, Roger, and Rachel built takes the outputs of project-level risk models, treats them as inputs into a program-level model, and produces a valid stochastic rollup without discarding the project-level work already done. The second structural problem the paper addresses is reserves. Programs typically maintain two types: project contingency held at the project level and management reserve held centrally. Without a model that shows the difference between summed project estimates and an integrated program estimate, there is no defensible basis for sizing the management reserve. Program managers set it by judgment. Executives approve confidence intervals they believe are accurate. The paper gives both a number with a method behind it. The framework also mandates conditions that programs should already have in place: a program-level risk owner, a steering committee, and a centrally managed budget. As Albert notes in the episode, the framework works best when those structures exist. When they do, the stochastic rollup model doubles as a summary schedule and a monthly reporting foundation. The window for influencing how this gets adopted as a standard is now. RISK-4852 is a paper, not yet a recommended practice. The AACE RP process requires practitioners to engage, test the framework on real programs, and contribute to the literature. Albert, Roger, and Rachel are presenting June 30 at the MGM Grand in Las Vegas and at the Safran Expo the following day. That is where the conversation starts. The gap is real and documented. The math behind "just add the P80s" is wrong. The window to build something better is open now. Another program cycle will close on a reserve number that was never correct if that window goes unused. Read the companion blog post: https://dokainish.com/insights/aace-r... Listener survey: https://forms.office.com/r/KFCi9aiENH Presented by Dokainish & Company www.dokainish.com The Risky Planner podcast delivers expert insights on project controls, capital project management, and strategic planning for today's complex business environment. Subscribe for regular episodes featuring industry leaders and practical advice.