Home → Troubleshooting → @RISK for Excel: Simulation → "Found invalid formula ... Continue without Smart Sensitivity Analysis?"
Applies to: @RISK 5.x–8.x
When I start a simulation, I get the error message
Found invalid formula ... Continue without Smart Sensitivity Analysis?
Or,
The reference '...' in '...' has an OFFSET or INDIRECT, whose precedents cannot be traced.
Continue without Smart Sensitivity Analysis?
Short answer: Click Yes, and the simulation will proceed. Tornado diagrams may show inputs that are not actually precedents of the output, if by chance they are significantly correlated with the output. If you want to avoid clicking Yes each time, keep reading.
Full answer:
Smart Sensitivity Analysis, referred to in the message, was introduced in @RISK 5.0. It makes tornado charts and other sensitivity measures ignore inputs that are not actually precedents of the output in question. You'll find a full description of this feature in the article Precedent Checking (Smart Sensitivity Analysis).
As @RISK traces back from your @RISK outputs to your distribution functions, along the way it may find an invalid Excel formula. "Invalid" doesn't necessarily mean the formula is invalid according to Excel's rules; it could mean a valid formula that Smart Sensitivity Analysis can't cope with. Go to the location given in the message (F5 is the keyboard shortcut), or use Ctrl+F to search for the function mentioned. Look at the formula. If it is actually invalid, fix it and re-run your simulation.
Smart Sensitivity Analysis cannot work with certain Excel functions, even in valid Excel formulas. Please see Precedent Checking for the list. If the message identifies one of these functions, do any of these:
If the formula is actually correct, and does not fall in any of those categories, please contact Technical Support.
What happens if I disable Smart Sensitivity Analysis, either in Simulation Settings or by clicking Yes in the message? Will my simulation results be wrong?
Your simulation results will be correct. Tornado graphs and other forms of sensitivity analysis may include one or more inputs that are significantly correlated with the output, even though those inputs are not precedents of the output. To exclude an inappropriate input, see Excluding an Input from the Sensitivity Tornado.
I got a similar message, but it mentions RiskMakeInput.
The RiskMakeInput function is treated as a special case, and the above techniques will not work for it. See "Found invalid formula ... Continue without detecting RiskMakeInput precedents properly?"
Last edited: 2018-03-27