KatRisk Releases FMKat

KatRisk LLC announced the official launch of its third product: FMKat. After SoloKat, our worldwide data and loss calculation platform for location level risk (flood and wind currently, more to come), and SpatialKat, the multi-peril probabilistic modeling platform, FMKat is a standalone sampling engine and financial model.

With FMKat, our customers can quickly iterate on financial model parameters when running their catastrophe models. FMKat allows users to import ground up loses by event-location-coverage from any catastrophe model and then apply the SpatialKat sampling and financial model engine. Core features of our FMKat implementation are:

  1. Location-aware sampling which enables coherent measures of risk (monotonicity, positive homogeneity, sub-additivity, translation invariance) in cat modeling which always produces identical results regardless of who runs the code on multiple operating systems. There will then be no artificial risk diversification when aggregating losses from locations to portfolios (or across portfolios).

  2. Improved loss convergence with repeatable sparse antithetic latin-hypercube sampling with 2 to 1000 samples. Our open implementation of location level uncertainty allows users to own their own risk.

  3. Flexible spatial kernel with uncertainty correlation between 0% and 100% in steps of 1% to ensure no drop-off in EP curve after event catalog maximum years. Location level EPs are not impacted by uncertainty correlation choices.

  4. Fastest run-time of any financial model in the industry with efficient in-memory architecture. Runs on Raspberry Pi and laptops to multi-core servers with close to 100% linear scaling based on fast I/O.

  5. KatRisk support with suite of pre- and post-processing scripts that can customize input and output, or optimize financial model parameters given any objective function. FMKat can be kicked-off from the command line, API or GUI.

SpatialKatModelDiagram_FM.jpg