Original Post Date: Tuesday, July 6, 2010 

Those words describe the triangular probability distribution used by FRISK in performing risk analysis in TruePlanning. FRISK requires two major assumptions on the part of the user. The first is that the combination or convolution of a number of triangular distributions results in a log normal distribution. The second is that there is correlation between cost objects. 

The triangular distribution is completely defined by three simple inputs: an optimistic value, a pessimistic value, and a most likely value. By eliciting information from engineers, I have found that they are much more willing to commit to a range for a certain parameter, rather than a point value. I then use the limits of the range for the optimistic value and the pessimistic value and then ask the question, within the range that you just gave, what is the most likely value? This then gives me all the information I need to satisfy the requirements of the triangular distribution.

Another unique feature of the triangular distribution is that the optimistic and most likely or pessimistic and most likely values can be set to the same value yielding a right triangle. The triangular distribution is the only one with this property.

So, the next time you perform a risk analysis using TruePlanning think of the lowly triangular distribution and how simple, yet powerful it really is.

John Long
Solutions Consultant, PRICE Systems