Your Project Behaves Like a Hurricane--Forecast It Like One
summary
Claims a concrete forecasting workflow: collect track start and end dates for every item, compute historical throughput, feed that into a Monte Carlo simulation (the demo runs 10,000 trials) and produce probabilistic dates — e.g. 50% by Dec 15, 85% by Dec 24, 95% by Dec 29 for a 100‑item backlog. Shows a live five‑minute group release plan and a running continuous forecasting dashboard used to update forecasts daily and drive scope/date decisions.
AI-written summary · about
chapters (10)
- 0:00Start
- 11:04On October 22, 2012, a storm formed in the southern Caribbean...
- 23:25What do we need to think probabilistically about?
- 23:40What does it even mean to think probabilistically?
- 25:06There are two problems with using an average to answer WWIBD
- 26:14A forecast should contain two parts: a range and a probability
- 26:49From Monty Hall To Monte Carlo
- 33:35Update Your Forecasts When You Get More Information
- 37:11Shorter Term Forecasts Are Better Than Longer Term Ones
- 39:07"Ultimate Kanban" Case Study at InfoQ
related talks
44mAgile metrics for predicting the future
30mSoftware Estimation Made Easy
16mPractically predictable – Forecasting delivery when nothing is certai
40mWhat are the odds? Transforming forecasts through probabilities
47mCreating efficient, accurate, software estimations
36mThe Development Metrics You Should Use (but Don’t)
43mPredictable Delivery
51mGet Predictability Regardless of Estimates