## What is the pitfall of Reserving Methods that use Triangles?

In various triangle reserving methods, loss information is "squeezed" away during the triangle construction or accumulation process. This becomes a substantial pitfall because rich information is degraded to poor information. The loss distribution may lose key information such as trend because of that "simplification". Thus, various methods such as linear or non-linear, deterministic (Chain Ladder) or stochastic, assumed distribution or bootstrapping, would still not be sufficient to yield quality estimates of pure IBNR and case development.

## Experiment

Let us do a simple experiment with Chain Ladder method on ultimate reported count, to see if we can obtain a peace of mind. Please note that, Reported Count triangle usually has a complexity level of 2 or 3, while Paid Loss triangle's complexity can go above 10 (applying payment lag, severity, trend, limit and deductible,correlation, etc). This example also illustrates how ReservePrism solves the puzzle.