Loss Trend Forecaster
Fit exponential/linear trend lines to historical frequency, severity, or pure premium and project forward with confidence bands.
Holt level smoothing (α)0.40
Holt trend smoothing (β)0.20
Projection: Holt vs. log-linear trend
| Forecast period | Holt linear trend | Log-linear fit |
|---|---|---|
| Year +1 | 12,312,316 | 12,596,078 |
| Year +2 | 12,839,330 | 13,302,973 |
| Year +3 | 13,366,344 | 14,049,540 |
Log-linear annual growth: 5.61% | R²: 0.996
Rate vs. trend
| Quantity | Value |
|---|---|
| Loss trend factor | 1.124 |
| Premium trend factor | 1.040 |
| Rate factor | 1.040 |
| Projected loss ratio | 67.5% |
Combined ratio
104.5%
Assumptions & limitations
- Holt's method requires at least 2 data points to initialize level/trend and assumes an additive, non-seasonal linear trend.
- The log-linear fit assumes constant multiplicative (exponential) annual growth; R² close to 1 indicates a good exponential trend fit.
- Rate-vs-trend and combined-ratio calculations assume trend/rate changes are geometric (compounded) and apply uniformly across the projection years.