A percentile is not a second selected ultimate. It is a framing device around the one you chose.

Mack and ODP bootstrap ranges describe process and parameter uncertainty under stated assumptions. They do not replace judgment about case adequacy or a priori bias.

Put the selected ultimate next to the 75th percentile and the adverse deviation threshold you care about. Then write why the selection sits where it sits.

How to use the range in the memo

Lead with the selection and reason. Add the range as context, not as the decision.

If the selection sits above the threshold, say what risk that absorbs. If it sits below, say what you are not loading for.

Avoid these mistakes:

• Treating the mean as the pick
• Ignoring that bootstrap paths share triangle noise
• Mixing unpaid and ultimate scales
• Skipping the assumption list
• Using one bad year to reject the whole model
• Hiding the selected point off the chart

Stochastic output belongs in the same production file as the deterministic methods so the actuary is not juggling tools.

numdog keeps Mack and bootstrap beside chain ladder and BF for that reason.

Ranges without a selection are decoration. A selection without a range is incomplete for many boards.

The range informs. The actuary still signs the number.

‍