Methods and settings
| Method / input | Behavior |
|---|---|
| Minimum / Mode / Maximum | Minimum ≤ mode ≤ maximum. |
| Shape λ | Advanced control; default 4, greater than 0 and at most 100. |
A weighted center within bounds
PERT uses the minimum, maximum and mode to construct a smooth bounded distribution. Larger λ concentrates more weight around the mode. Before additional conditioning, the mean is (minimum + λ × mode + maximum) / (λ + 2). The default λ = 4 weights the mode four times in this expression; it is not a confidence level.
Synthetic configuration
Minimum 80 EURk, mode 100 EURk, maximum 140 EURk and λ = 4 imply a mean of about 103.33 EURk before further bounds. This differs from the Triangular mean for the same three points. Review the shape and resulting statistics rather than treating both families as interchangeable.
Do not overstate precision
Equal endpoints produce a constant distribution. Invalid ordering or shape values must be corrected. A high λ narrows uncertainty because you instructed the model to do so; it does not demonstrate stronger evidence. Additional bounds condition the distribution, and integer inputs require Discrete or Empirical alternatives.
Documents the current implemented behavior. Numerical examples are synthetic, not market assumptions.