Cyber Risk Quantification Platform
An actuarial simulator that turns raw threat data into a number a finance team can act on: expected loss, tail risk, and a price.
The problem
Cyber risk usually gets described, not quantified. A red, amber, green rating tells a board nothing about how much a bad year actually costs, and a single average loss hides the tail, which is the part that matters. Insurers and risk teams need a distribution, not a traffic light.
What I built
An actuarial Monte Carlo simulator that models the two things separately: how often loss events happen and how bad each one is. It uses a Poisson-Gamma compound distribution, Poisson for event frequency and Gamma for severity, and runs thousands of simulated years to build the full loss distribution.
- Expected loss and tail risk fall out of the distribution, so you can talk about the average year and the bad year in the same language.
- Those figures translate directly into financially interpretable risk scores and insurance pricing recommendations.
- A full-stack dashboard makes the inputs and the resulting distribution explorable rather than buried in a notebook.
Why it matters
Modelling frequency and severity separately is what lets the model price the tail honestly. It moves cyber risk from a qualitative colour to a number a finance team can underwrite against.