catastrophe-modeling

This white paper proposes using LSTM neural networks to detect structural breaks in property insurance loss reserving caused by climate-driven catastrophes, aiming to improve accuracy by 15–20% over traditional methods like Chain Ladder.