A risk model in Risk Kit does not just deliver a forecast. It can also be updated and checked. This is an important difference from a purely expert-based assessment. The 2026 World Cup illustrates this nicely: before the tournament, our Monte Carlo model was based essentially on the teams' Elo strengths. After 36 group-stage matches had been played, we could simply update the model: matches already played are carried over as fixed facts in the tournament path, new Elo values flow in, and Risk Kit simulates the title chances again. The first advantage of an explicit model: it can be easily updated and re-evaluated at any time. The second advantage is just as important: the assumptions can be checked. For the World Cup model, we validated three central assumptions after the halfway point of the group stage. The average number of goals is 3.03, somewhat above the model assumption of 2.67 goals per game, but still within a statistically plausible random range. The Poisson assumption for goal counts shows only slight overdispersion and is not contradicted by the data so far. A dependency between the goal counts of both teams is also not detectable so far. 36 games are not yet a large sample. But the model's assumptions are transparent from the outset and are already verifiable and open to discussion, long before the World Cup is over. With a purely expert-based assessment, this is much harder. You can see afterward whether a result occurred or not, but it remains unclear which assumptions were behind it, which data supported them, and whether the outcome was still chance or already a sign of an unsound assessment. Risk models built with Risk Kit make assumptions visible, updatable, and capable of being validated. This is crucial for Enterprise Risk Management: risk assessments must not only be produced. They must be traceable enough that management trusts them. Here are the links to the World Cup model. Halfway through the group stage: https://lnkd.in/eVwQJmxF Before the start of the World Cup: https://lnkd.in/eCihyAzt
This post was originally published on LinkedIn. View and join the discussion there