A global reinsurer’s flow business was losing an estimated $1 million annually through sub-optimal pricing. With pricing cycles running every two weeks, the actuarial team was spending hours manually gathering competitor rates across multiple markets before each cycle.
By the time rates were set, competitors had already moved. The team was stuck in a reactive loop - choosing between margin preservation and market share without enough data to support either decision.
What we did
We built two ML models that work together. The first predicts competitor crediting rates two weeks ahead, drawing on historical rate data, bond yields, equity indices, treasury rates, and FX movements. The second forecasts how cedant sales volumes respond to pricing changes - so the team can see not just where competitors are heading, but what happens to volume at different price points.
Instead of always being the lowest rate by an unnecessary margin, the models enable “just ahead” pricing - setting rates to capture volume while preserving yield, based on where competitors are actually heading. The model also accounts for capital management constraints - sometimes being the best rate isn’t desirable if you’re approaching reserve limits.
Pricing meetings shifted from reactive discussions based on stale data to proactive scenario planning with forward-looking intelligence.
Why it matters
In flow reinsurance, the margin between winning and losing a deal is often just a few basis points - or a few hours in the pricing cycle. For Bermuda reinsurers in concentrated markets, that kind of timing advantage is hard to replicate once a competitor has it.