Home → Troubleshooting → @RISK for Excel: Simulation → "The correlation matrix at ... is not self-consistent."
Applies to: @RISK 4 and newer
When my simulation starts, I get this message:
Warning
The correlation matrix at ... is not self-consistent. @RISK can generate the closest self-consistent matrix. OK generates a corrected matrix and continues, Cancel stops the simulation.
It's tempting to click OK, but that probably won't generate the matrix that best represents your correlations.
If you have @RISK 5.5 or newer, please see How @RISK Adjusts an Invalid Correlation Matrix for information on using an adjustment weights matrix.
If you have @RISK 5.0 or older, or you just want to adjust the matrix yourself, see How @RISK Tests a Correlation Matrix for Validity.
In the message I got, the location is "(In)Depc_matrix_sourcing", but Excel says that's not even a valid name.
You're using the legacy functions, RiskDepC and RiskIndepC. At the start of simulation, @RISK builds a correlation matrix from those functions, but assigns zero to correlations among the dependent variables; see Correlation Matrix Equivalent to RiskIndepC and RiskDepC in the attached workbook. It's not unusual for that matrix to be inconsistent, as in Example 1 in the workbook. If you get the above message and click OK, the simulated correlations may be very different from what you specified. See Simulated Correlations if You Click OK in the attached workbook. @RISK leaves the matrix it actually used in CORRMAT.MTX in your temp folder, but you can see the problem just from looking at the simulated correlations.
Recommendation: Assuming you have @RISK 5.5 or newer, please see How and Why to Switch from RiskDepC to RiskCorrmat. The technique in that article will lead @RISK to preserve your specified correlations to the extent possible, even if the matrix is not self-consistent.
Last edited: 2017-08-08