How would you create a probabilistic forecast for a backlog?

This question tests your ability to use flow metrics for probabilistic forecasting, moving beyond single-point estimates. A strong answer outlines using historical weekly throughput data to run a Monte Carlo simulation. This involves thousands of random samples to model future performance. The result is a range of delivery dates tied to confidence levels (e.g., an 85% chance of finishing by Week 10). A major red flag is using averages for forecasting, which hides variability and gives a false sense of precision.