@RISK: Same Correlation Coefficients for Several Groups of Inputs

Use the optional "instance" argument in RiskCorrmat() to correlate inputs within groups without correlating across groups.

Jan 14, 2026
Knowledge

Title

Answer

Applies to: @RISK 5.x and Up

Correlating Multiple Groups of Inputs in @RISK

Problem

You want multiple groups of inputs (e.g., A, B, C and D, E, F) to be correlated within each group using the same correlation coefficients, but not correlated between groups. By default, @RISK correlates all inputs together.

Solution — Using Instances in RiskCorrmat

@RISK lets you assign a different instance to each group of correlated inputs. Inputs within a group share the same instance, and inputs in different groups use different instances.

Method A — Point-and-Click:

Highlight the first group of inputs. Right-click and select @RISK » Define Correlations or use the ribbon icon.
Enter the correlation coefficients in the matrix.
Check the matrix consistency.
Set the matrix location and optionally name it.
Click Rename Instance and assign a unique identifier for this group.
For additional groups using the same matrix, click Create New Instance, assign a unique identifier, and select the inputs.
Repeat for all groups and click OK.

Shortcut for Contiguous Arrays:

Use Create Correlated Time Series in the Define Correlations dialog, select the rectangle containing all groups, choose correlation by rows or columns, set coefficients, check consistency, name the matrix, and click OK.

Method B — Formula Editing:

Create a named correlation matrix for your group of inputs.
For each input, include the RiskCorrmat function in the distribution formula:
RiskDistribution(args, RiskCorrmat(matrixname, inputnumber, instance))
Use a cell reference for the instance so that all inputs in a group share the same instance.
Highlight the first group and drag the fill handle to replicate the formula for additional groups, adjusting the instance reference as needed.

After running the simulation, the CorrelationAudit_Report worksheet shows that inputs within each group are correlated as specified, and there is no correlation between groups.

Attachments:
KB0073_CorrelationGroups.xlsx

URL Name

same-correlation-coefficients-for-several-groups-of-inputs