Man Group analyzes accuracy of Atlantic storm forecasts

Investment firm Man Group PLC has released an analysis evaluating the accuracy of professional Atlantic hurricane season forecasts. The study compares predictions from entities like NOAA, the Met Office, and Tropical Storm Risk (TSR) against a "naive" forecast derived from the average storm counts of the preceding ten years.

The analysis introduces a "skill" metric that measures the error of forecasts against the error of the naive prediction. Results show a mixed performance, with over 40% of forecasts performing worse than the naive baseline when considering all data. Even over longer periods and for specific metrics such as the number of hurricanes and major hurricanes, the calculated "skill" often remained below 20 percent.

These findings are significant for the insurance and reinsurance sectors, which may use seasonal storm forecasts to inform pricing for reinsurance contracts. While forecasts offer context for investor discussions, the analysis underscores their potentially limited predictive power.

Man Group highlights the challenges in conducting like-for-like comparisons and the importance of clear data definitions. Although simple, the "naive" forecast serves as a crucial benchmark for assessing professional predictions. The results suggest that market participants and insurers should approach the precision of storm forecasts and their impact on reinsurance pricing with critical consideration.