• Location Risk Intelligence turns underlying probabilistic and other statistical climate risk models into deterministic metrics such as hazard zones, Climate Expected Loss, Risk Scores and 1 in 100 Year Damage, so that risk managers can work with clear numbers instead of raw simulations.
• The article explains why some perils do not have full frequency intensity curves, why Location Risk Intelligence may disagree with competitor or governmental flood maps, and how Risk Score classes are defined and skewed, including the role of wildfire in the Overall Risk Score.
• It shows how to interpret apparent contradictions between Climate Expected Loss and Risk Scores at a single site and how to use both metrics together to steer underwriting, lending and portfolio resilience decisions more confidently.

The short answer is that some hazard models in Location Risk Intelligence use probabilistic methods in the background, but all outputs you see in the platform are deterministic.

On the user side, Location Risk Intelligence always serves pre calculated, deterministic metrics for each location and peril, for example hazard zones, Climate Expected Loss (CEL), Risk Scores and 1 in 100 Year Damage. These metrics are extracted from underlying hazard and loss curves, so that you can work efficiently with location-specific numbers and easily compare locations without having to run heavy simulations yourself.

This means you get the benefit of probabilistic and statistical science, packaged into deterministic outputs that are practical for portfolio screening, underwriting, lending, site selection and disclosure.

Some perils simply do not offer enough clean data for a robust hazard curve with multiple return periods. For hazards such as hail, flash flood or certain wildfire models, Risk Management Partners often faces two constraints:

• Sparse observations. There are not enough reliable historical events at each location to fit a meaningful frequency intensity distribution.
• High model complexity. Fully dynamic hydrological or fire behaviour models would be computationally extreme at global scale.

In those cases, Location Risk Intelligence still gives you a decision ready signal by modelling discrete risk classes, based on how often climate conditions that favour the event occur, for example intense rainfall plus steep slopes for flash floods.

You can think of this like an election poll in a small town. With limited responses, you can still see which way sentiment leans, but you would not claim precise percentages. The same principle applies to these perils. The classifications are useful for risk ranking and screening, even when a full return period curve is not available.

We will never claim that our model is universally “the best”, because the full methodologies and data of competitors are not visible. Like any climate model, ours and others are approximations of a complex reality, each with its own strengths and weaknesses. What we can do is explain why you can have confidence in Location Risk Intelligence:

• Location Risk Intelligence uses peer reviewed data from recognised providers and institutions.
• Methods are anchored in the scientific literature, not in opaque black boxes.
• Models explicitly explore plausible extremes beyond the historic record, using statistical extrapolation and stochastic event generation.

Let’s say, you compare Location Risk Intelligence to a competitor or governmental map. Location Risk Intelligence shows river flood risk at a site. The other map indicates no risk. When two maps disagree, it simply shows that different modelling approaches and weightings of climate drivers are at work. This is the moment to ask: which model reflects physical reality and changing climate drivers more consistently for my decision?

Risk Score intervals are set in line with internal Munich Re underwriting guidelines for risk and expected loss levels. The exact loss rates and thresholds are confidential, but the distribution is highly skewed:

• Less than 5 percent of locations have a Risk Score above 100.
• Less than 0.4 percent of locations have a Risk Score above 200.

So if a hospital, data centre or logistics hub appears with a Risk Score above 100, it is already among the more exposed locations in the global distribution, even if its Climate Expected Loss is moderate.

Climate Expected Loss is location specific. It uses the hazard curve and vulnerability information at that precise coordinate.

Risk Scores are hazard zone specific. The score reflects the average annual loss of all locations globally that sit in the same hazard zone. If your asset is in a relatively benign part of Zone 50, its Climate Expected Loss can be near zero because flood depths at that specific coordinate remain shallow or damaging events are rare, while the zone-level Risk Score is high because many other locations in Zone 50 experience deeper, more damaging floods.

For portfolio steering, the combination is powerful – but if you have access to Climate Expected Loss values, you should treat them as your primary decision metric. CEL reflects the location-specific hazard profile and gives you a more precise, interpretable signal, typically expressed as a permille of the asset value. Risk Scores complement this by covering all four perils in a single index, including wildfire, and by giving you a quick way to rank and segment large portfolios where detailed CEL information is not yet available or not needed at every step.