Diligence pack

Everything your risk team will ask, answered in writing.

FloodGate is evaluated asynchronously by design. This page is the first meeting — methodology, validation, data provenance and licensing, in one place.
0 deviationsPython engine vs the published research workbooks (951 loans + 37,966 assets × 3 scenarios)
99.9% within 5 cmautomated depth extraction vs the paper's manual GIS work (r = 0.999)
689 layersfull WRI Aqueduct v2 catalogue, re-extractable in ~90 minutes
100% open dataevery input CC-BY or public domain; every output traceable to a CSV row

Methodology

Expected damage follows Eq. 3 of Assab (2025), Journal of Climate Finance 11:100066: probability-weighted damage across return periods, depth-damage functions from Huizinga et al. (2017). The financial translation applies the paper's regression on 951 project-finance loans matched to Moody's data: +99 bps probability of default per one standard deviation of expected damage; enforced flood protection standards −4%. Full papers: 10.1016/j.jclimf.2025.100066 · 10.3390/jrfm17040133.

Data provenance & licences

Architecture & auditability

The platform is a static dataset plus client-side computation: no black box between data and screen. Every number displayed can be traced to a row of the published results database and recomputed from the cited equations. Hazard data refreshes are versioned; analyses record the data version they were computed on.

Known limitations (stated, not hidden)

Security & privacy

The free product stores registration locally and requires acceptance of the usage policy. Premium portfolio data is private to the customer's profile by default and never enters the public database without explicit consent.

Questions

Written answers within two business days: a.assab@sms.ed.ac.uk.