Book

Description

Reinsurance pricing in India still runs largely on judgment, spreadsheets, and years of accumulated instinct. This book asks a simple question: what happens when you give underwriters a better number to work with? It walks through a seven-layer underwriting pipeline built for the Indian general reinsurance market, built around something called the Unified Risk Index — a single score that pulls together machine learning risk predictions and Solvency II capital requirements into one figure an underwriter can actually use. Tested against 250 simulated reinsurance deals, the approach cut loss ratios by close to 9 percentage points, working out to roughly Rs. 95 crore in losses that could have been avoided. The chapters cover how the system takes in data, scores risk, catches fraud, factors in weather exposure, calculates regulatory capital, and learns from its own past decisions. It's written for people who already know the industry — actuaries, underwriters, reinsurance analysts — and who are curious whether machine learning has a real, practical role in how risk gets priced, not just a theoretical one. This isn't a manifesto about AI replacing underwriters. It's closer to a field report: here's what was built, here's how it performed, here's where it might go next.