Moving a model from @RISK, ModelRisk or Analytic Solver

Guide · Updated October 2026

xellstorm opens a workbook built for @RISK, ModelRisk or Analytic Solver without the add-in, reads 164 of their distribution functions, and turns each such formula into an input whose parameters stay linked to your cells. Before you run, it lists what it converted and every formula it left, with the reason.

The 164 functions are 53 of @RISK, 62 of ModelRisk and 49 of Analytic Solver. It also reads 65 forms that give a distribution by its percentiles, along with truncation, output markers, correlation matrices and multi-simulation tables.

What the import finds in a workbook written for @RISK

The example is the building estimate of the project contingency example, written the way a model built for @RISK would be: each cost item is =RiskPert(min, most likely, max, RiskName(label)) on the cells of its row, each risk event is a RiskBernoulli on its probability, the total cost and the overrun are marked with RiskOutput, Structure and MEP are correlated through RiskCorrmat, three budgets sit in one RiskSimtable, and a cell reports RiskMean of the total. A second sheet, “Demonstrations”, holds five separate formulas: two arithmetic forms that do convert and three formulas that do not. We made the file with a script; xellstorm does not need @RISK to read it, and neither do you.

In the app, Import from workbook on the Distributions step lists the formulas it found (the command line’s xellstorm import prints the same list). For this workbook it converts 12 inputs: the 6 cost items, the 4 risk events and 2 arithmetic demonstrations. The cost items and risk events keep their parameters linked to the cells, so editing a min or a probability in the workbook changes the next run, and RiskName(A3) names each input after the text in its row. The two that convert are a Normal draw multiplied by a factor, and one whose mean is calculated from another cell. Neither is used by the estimate’s outputs.

The same import reads the rest of the model:

• Outputs. The cells marked RiskOutput("Total cost") and RiskOutput("Over budget") become outputs with those names. The marker adds 0, as it does in @RISK, so the cells calculate.
• Correlation. RiskCorrmat(Correlation,1) and RiskCorrmat(Correlation,2) put Structure and MEP in the matrix under the name Correlation, a rank correlation of 0.6 (only the lower triangle is filled in; a blank cell is read from its mirror across the diagonal). The value is read from the matrix as the workbook holds it when you import, and the import says so.
• Simulations. RiskSimtable({2900,3000,3100}) gives the budget one value per simulation in @RISK. xellstorm holds the budget at 2,900 in the base run and adds the other two as scenarios, which draw the same random numbers as the base run, so they differ only in the budget.
• Statistics. RiskMean, like the other functions that report results back into a cell, is not a distribution, so it is not listed as unconverted. It has no value outside @RISK (the compatibility check names the cell); the mean is in xellstorm’s results.

Formulas that do not convert are listed too. All 3 on the second sheet come with the reason the import gives, and each has a way forward:

Running the converted model

Run with the imported setup and 5,000 trials, the average total cost is 2,784.7 (USD thousands), and the exact mean of the same distributions, the PERT means (min + 4 × most likely + max) / 6 plus each risk’s probability times its impact, is 2,784.7; the standard error of the simulated mean is 1.7. Structure and MEP reach a rank correlation of 0.60. The three budgets of the RiskSimtable, run as the base and two scenarios:

A larger budget is exceeded less often, and because the scenarios draw the same numbers as the base run, the differences between the rows come from the budget alone, not from sampling.

Will the results match the add-in?

The model is the same; the random numbers are not. Each rule of the import maps a function to the distribution it describes, with the parameters translated as the vendor documents them, and is tested against the equivalent scipy distribution; the script that makes the test data also checks each mapping’s mean against the vendor’s documented formula. So once every distribution of a workbook converts, its means and percentiles differ from the add-in’s by sampling noise, which shrinks as you run more trials: the standard error above is the size to expect for the mean.

Correlations are rank correlations, as @RISK and Analytic Solver define them, and xellstorm reaches the rank correlation asked for. How the draws are paired to reach it is each tool’s own, so a correlated model can differ a little more in the tails than sampling noise alone explains. The workbook itself is not changed: xellstorm reads the file, and your model still opens in the add-in.

What converts

• Distributions and arithmetic arguments. A distribution can stand alone, or have one positive multiplier followed by at most one added or subtracted offset: RiskNormal(B1*1.1,B2)*C1+5 and 5+C1*RiskNormal(B1,B2) convert. Arguments, multipliers and offsets can use numbers, single-cell references, parentheses and +, −, *, /. Defined names that refer to one cell become its address. These expressions stay linked and are evaluated after each scenario’s fixed cells are applied; every reference must be deterministic. Functions marked † in the table still convert with numbers only.
• Percentile forms. @RISK’s …Alt and …AltD functions and Analytic Solver’s …Alt functions (those in the table below) give a distribution by its percentiles; xellstorm solves for the distribution exactly when you run and reports an error when no distribution, or more than one, fits them.
• Truncation and shift. RiskTruncate, RiskTruncate2, RiskTruncateP, RiskShift, VoseXBounds, VosePBounds, VoseShift, PsiTruncate, PsiTruncateP and PsiShift, applied in the order each add-in documents. Any multiplier and offset outside the distribution apply afterward and are shown separately in the input’s settings.
• Names and locks. RiskName, PsiName and VoseInput name the input; RiskLock holds it at one value. The catalog also lists properties that are ignored (Static, Units, Category, Collect, Base, Seed, IsDiscrete, IsDate, Fit, FitInfo, Certify, BaseCase, Library and SixSigma, with the vendor prefix). RiskSeed produces a warning: its random-stream settings are not preserved, and xellstorm’s simulation seed and input order control its draws.
• Outputs. RiskOutput, VoseOutput, PsiOutput and PsiSimOutput mark outputs; in Analytic Solver a statistic such as PsiMean(B1) also makes B1 an output, and so it does here.
• Correlations and simulations. RiskCorrmat, RiskDepC with RiskIndepC, PsiCorrMatrix, PsiCorrDepen with PsiCorrIndep; RiskSimtable, VoseSimTable and PsiSimParam.
• Excel’s own formulas. NORM.INV(RAND(), mean, sd), RANDBETWEEN, IF(RAND() < p, 1, 0) and 18 such forms in all, listed below.

What does not convert

Multiple distributions in one cell, a distribution inside another function or in a denominator, direct division of a distribution, and chains that require changing the order of arithmetic do not convert. For example, (RiskNormal(0,1)+5)*2 and RiskNormal(0,1)*2*3 are left with a reason. RiskNormal(0,1)*(1/C1) does convert when C1 gives a finite positive multiplier, because the formula explicitly calculates the reciprocal before multiplying. Multipliers that work out to zero or less, arguments that call functions or depend on random cells, functions with no exact equivalent (compound distributions, time-series functions, RiskPertAlt and the other two-shape percentile forms), ModelRisk’s U argument and copulas, and properties not in the table also remain unsupported. An unsupported add-in formula cannot be calculated by the engine; the compatibility check names it.

Supported functions

The tables are generated from the importer’s own rule list, so they are exactly what it converts. “Percentiles” marks the forms that give a distribution by its percentiles; † marks the functions that convert with numbers only. Under each distribution is its name in xellstorm specs.

Questions

@RISK, ModelRisk and Analytic Solver are trademarks of their respective owners. xellstorm is not affiliated with them; their function names appear here only to say what xellstorm reads.

Related

• Project cost and schedule

  How much contingency does a construction project need?

  Worked example: a building estimate’s most likely costs are exceeded in 94.2% of simulated outcomes. Size the construction cost contingency with Monte Carlo.
• Guide

  Monte Carlo simulation in Excel, with or without add-ins

  Three ways to run a Monte Carlo simulation on an Excel model: RAND() and a data table, an add-in, or a browser tool with no add-in. PERT formula included.
• Guide

  PERT distribution in Excel: formula, mean and when to use it

  A PERT is a beta distribution from min to max with mean (min + 4 × likely + max)/6. The Excel BETA.INV formula, and PERT vs triangular.

xellstorm is a browser-based Monte Carlo simulation tool for Excel models: no add-in, and the workbook never leaves your computer.