Here is a brief description of the artificial intelligence methods including statistical optimisation techniques and pragmatic solutions which the IBNR Robot combines to estimate losss reserves and its range.

1. Outlier Truncation

Data input beyond the acceptable threshold is truncated. This reduces the influence of unusually large or small values due to unusual claims events or data handling errors on the results.

2. Runs Test

Runs Test is applied to observed development factors and decay factors to arrive at an unbiased selected link ratio development factors which credibility, relevance and homogeneity is optimised.

3. Significance of Correlation Coefficient

Correlation coefficients between methods, calendar years observations and classes of business which are statistically significant and pragmatically acceptable are applied to the variance covariance matrix through fuzzy inclusion. 

4. Method of Lagrange Multiplier

Method of Lagrange Multiplier is applied to the estimated variance to arrive at the optimised seed loss ratios for Bornhuetter-Ferguson method as well as the optimised loss reserves estimate.

5. Method of Moments

Method of Moments is applied to estimated mean and variance to derive the loss reserves range for each line of business as well as those for the aggregate company.

6. Variance Adjustment

The variance is pragmatically adjusted to account for uncertainties within adjustments applied throughout the calculations.

Please reload

Automating Actuarial Work

IBNR Robot