Mark Simms
marksimms at verizon.net
Thu Dec 22 17:06:24 CST 2011
Omigosh, this one is a great story for the EUSprig group...they track all business-related spreadsheet disasters that have a financial consequence. So let me get this straight, their scaling went like 1.0, 1.5, 2.0, etc ? So that required a special rounding function....rounding to the nearest 0.5. So 1.26 would go to 1.5, 1.75 would go to 2.0....correct ? The issue was a series of 4 questions would return a value based on a risk matrix. So you would have a value like 1.99 or 1.67 and it would all be regrouped as a 1. They were wondering why their rating engine were returning screwy results (it was basically a risk scale so anything with a risk rating >= 1.5 should have been priced using risk band 2 for example). Given that this data was used to price customer quotes based on their risk profiles they were underquoting consistently until I spotted the error