# Estimating with Uncertainty

In Agemba it is encouraged (but not mandatory) to take uncertainties into account when estimating effort and value. This is done by setting not only one single estimate for the effort and value, but setting both an expected and a worst-case estimate.

The expected estimate is what is considered as a realistic estimate with 50% certainty (it should not be confused with an optimistic best-case estimate). The worst-case estimate represents a realistic estimate with a 90% certainty. By considering not only the expected estimate, but also the worst-case, you force yourself to consider different scenarios and avoid the mistaken perception that estimates are certain and precise.

The uncertainty in percentage is derived as the difference between the expected and worst-case estimate, divided by the expected estimate. The higher the difference is compared with the expected estimate, the higher the uncertainty.

The worst-case effort estimate will always be higher than the expected estimate, hence the effort uncertainty can increase to more than 100%. The worst-case value estimate will always be lower than the expected value estimate, hence the value uncertainty will always be lower than 100%.

#### Accumulated Estimates

The accumulated expected estimate is the simple sum of the expected estimate of all children.

However, deriving the accumulated worst-case estimate is not as simple. It depends on whether you assume all the children (cards found under any major card) to be mutually independent (i.e. if one story takes more time, this does not necessarily affect the others) or if you assume the children to be mutually dependent (i.e. if one story takes more time, so will all probably). Based on this assumption, the worst-case estimate can be derived using either of the two methods:

1. Root-sum-squared for mutually independent children – less than a simple sum of worst-case estimates: expected +/- √(∑(worst – expected)2)
2. Simple sum of worst-case estimates for mutually dependent children