This review compares Monte Carlo simulation add-ins for Excel. If you are still mapping the wider category — what risk analysis software does, the four types, and how to choose between them — start with our guide to risk analysis software.
What you may like:
What you may NOT like:
Price sources (3-year costs are approximate list prices for one user, before negotiated discounts): Analytic Solver Simulation is listed at $2,500/year on solver.com (retrieved August 2026); Lumivero’s web shop lists @RISK Professional at €2,900/year and @RISK Industrial at €3,500/year (shop.lumivero.com, retrieved August 2026, shown in local currency) — converted to USD here at the same EUR→USD factor used for ModelRisk prices on this site; Oracle does not publish Crystal Ball pricing and Capterra lists it as contact-vendor (retrieved August 2026); RiskAMP Professional is $279 one-time with renewals at half price per riskamp.com (retrieved August 2026); ModelRisk prices come from our published price list.
If budget is tight, RiskAMP is the cheapest option, but it falls short of professional standards. Oracle Crystal Ball is showing its age - development has stalled, the interface feels dated, and its feature set lags behind the competition; Oracle no longer even publishes a price for it. Our detailed comparison is in ModelRisk vs Oracle Crystal Ball.
For serious work, the real contest is between ModelRisk, Analytic Solver, and @RISK. ModelRisk offers the widest range of technical features at the lowest professional-grade price. If you need stochastic optimisation today, Analytic Solver is the one to look at - though that gap is closing: ModelRisk Optimizer is built and will be released in the near future. Frontline Solvers also sells a separate Data Mining tool alongside Analytic Solver, though buying both gets expensive quickly.
@RISK remains the most expensive option: Lumivero's web shop lists @RISK Industrial at €3,500 per year (shop.lumivero.com, retrieved August 2026) - more than twice ModelRisk's €1,550 published maximum - for a narrower technical feature set. The main reasons to pick @RISK are if you already hold a perpetual licence or need to keep working with existing @RISK models. See our detailed @RISK pricing breakdown.
Spreadsheets are everywhere when it comes to estimating things like project costs, NPVs, or sales forecasts. Usually these models give you a single number. The trouble is, you have no idea how confident you should be in that number.
A risk analysis add-in swaps out fixed inputs for probability distributions - functions that generate random values reflecting the uncertainty around each assumption. You mark certain cells as outputs, run thousands of recalculations, and get a full picture of what might happen and how likely each outcome is. That process is called Monte Carlo simulation, and Excel turns out to be a natural fit for it because the model logic is already there in your spreadsheet. This article explains Monte Carlo simulation in more detail.
Most people pick one Monte Carlo add-in and stick with it for years without checking what else is out there. These tools are not cheap, so it is worth stepping back occasionally and seeing how the alternatives stack up. We ran all five products side by side to give you an honest comparison.
Think about how complex your models actually are. If you are just summing costs, any add-in will do. But if you are an engineer who needs Weibull distributions, an epidemiologist working with hypergeometric models, or a financial analyst fitting log-normals, you need software that has them. The same goes for correlation structures and time series - check the tool supports what you need before buying.
Speed matters less than it used to, but it still matters in some cases. Banking capital allocation models, for example, may need 500,000+ samples to pin down tail probabilities. Stochastic optimisation requires many repeated runs. In those situations, a slow simulator wastes real time.
Some models need more than basic distributions. Convolution (summing distributions in a single cell), extreme value analysis, and reverse convolution (dividing one distribution by another) can save you from building awkward workarounds. Not every add-in offers these.
You will spend a lot of time looking at distribution plots and results windows, so the interface matters. At minimum, you want clear visualisations of probability distributions. Sliders and interactive controls that let you explore probabilities on the fly make a real difference in day-to-day use.
All five products use the Mersenne Twister RNG, which is standard across the industry. What matters more is how each tool turns those random numbers into valid distribution samples. Seed control for reproducibility and scenario tracking for rare outcomes are worth checking. Stratified sampling gets talked up a lot, but in practice it adds little unless your model has very few distributions.
If your organisation already uses a particular tool, there is an obvious pull to stay with it for compatibility. That said, the cost savings from switching can be substantial, especially across large teams. Several tools also include model converters that ease the transition.
Excel updates break add-ins more often than you would expect, and IT security policies can block installations or updates without warning. Before committing, try contacting each vendor's support team with a question and see how quickly and helpfully they respond.
Good example models save hours of trial and error. Some vendors also offer online courses, consulting, or on-demand training, which helps if your team is new to quantitative risk analysis.
At some point, your results need to go into a report or presentation. Tools that let you export charts and tables directly, and re-generate them when assumptions change, save a lot of copy-paste work - especially when someone changes an input five minutes before a meeting.
Most of these products offer free trials, typically 15 days - just enough time to build a test model and see how the tool feels. The exception is Oracle Crystal Ball: Capterra’s listing states that a free trial is not available (retrieved August 2026). ModelRisk is the only one that drops down to a free basic edition after the trial expires, so you can keep using it with reduced features rather than losing access entirely.
More organisations are moving to cloud-based Excel (Microsoft 365 online). It is worth asking each vendor whether their add-in works in that environment and what their roadmap looks like. Buying a desktop-only tool when your company is migrating to the cloud could be a problem.
Prices vary enormously across these products. The detailed cost comparison at the end of this article breaks down what you pay over one, two, and three years, plus value-for-money metrics like functions per dollar.
Products are listed alphabetically.
Made by Frontline Solvers, Analytic Solver combines Monte Carlo simulation with optimisation. The standout feature is stochastic optimisation: you define decision variables and the tool optimises against stochastic outputs like risk-adjusted profitability.
Monte Carlo simulation is only part of the package. The simulation engine is fast - Frontline knows Excel’s calculation engine inside out, and it shows. The software also includes AI-driven assistance and web publishing, though the Monte Carlo feature set itself is narrower than some competitors.
Crystal Ball was one of the first Monte Carlo tools for spreadsheets, originally built by Decisioneering, Inc. Oracle bought it years ago and has done very little with it since. Rather than placing distribution functions directly in cells, Crystal Ball overlays uncertainty on top of existing values, which feels clunky compared to how modern tools work. For a full head-to-head, see ModelRisk vs Oracle Crystal Ball.
ModelRisk was built specifically to fill the gaps left by earlier tools. It introduced time series modelling, copulas, and thorough distribution fitting to Excel-based risk analysis. It has the largest function library of any product in this review, and its dedicated Results Viewer makes it straightforward to analyse and present simulation output without leaving the Excel environment.
@RISK has been around longer than most of its competitors, and that head start gave it a large user base. Feature-wise it covers much of the same ground as ModelRisk, though it lacks some of the more specialised technical functions. The main sticking point remains price: Lumivero's web shop lists @RISK Industrial at €3,500 per year (retrieved August 2026), more than twice ModelRisk's €1,550 maximum for a narrower technical feature set. If you are weighing up a switch, read our migration guide.
RiskAMP is the cheapest option by a wide margin, but you get what you pay for. The feature set is minimal and the interface is bare-bones. It is fine for learning the basics of Monte Carlo simulation, but it is not really built for professional work.
We picked a few features where the differences between products are most visible.
Distribution fitting is a good test of each product's technical depth and interface design. We fed the same dataset into all four tools (RiskAMP does not have this feature) to see how they handle it.
One thing to watch for: some tools will happily fit a Uniform distribution to your data, which just takes the observed min and max as parameters. That implies the variable could never fall outside the range you have already seen - a dangerous assumption, especially with small samples. Good software should steer you away from that.
Parameter stability is another issue. Fit a Normal to three data points and the estimated mean and standard deviation will jump around wildly when you add a fourth observation. With fifty points, they barely move. Parametric bootstrapping accounts for this uncertainty, and not all products support it.
Finally, a distribution with four parameters will always fit data better than one with two parameters, simply because it has more flexibility. The Akaike Information Criterion (AIC) adjusts for this. ModelRisk was the first risk analysis add-in to use AIC; @RISK and Analytic Solver adopted it afterwards.
Analytic Solver's fitting algorithm works well, but the resulting function cannot reference the source data directly. If you later update your data, you have to re-fit manually. It also does not account for parameter uncertainty in the model.
Crystal Ball still uses older goodness-of-fit statistics (Anderson-Darling, Chi-Squared) rather than Information Criteria. In our test, this caused a four-parameter Beta distribution to rank above the Normal that actually generated the data - a classic overfitting problem. Like Analytic Solver, it cannot reference source data in the function and ignores parameter uncertainty.
ModelRisk uses AIC and supports Likelihood Ratio (LR) comparisons for ranking fits. The fitted function can reference source data directly, so the model updates automatically when data changes. It is also the only tool that incorporates parameter uncertainty into the simulation itself - visible here as the spread of fit lines. A fitted Normal linked to the data looks like this:
=VoseNormalFit(C3:C423)
@RISK's fitting algorithm is decent, though it is not entirely consistent in which criteria it applies across different distributions (note the Uniform result). It can reference source data in the function. Parameter uncertainty shows up in the fitting interface but is not carried through into the actual simulation, which rather misses the point. The mode estimate it displays has no real statistical meaning and should be ignored. A fitted Normal looks like this:
=RiskNormal(2.1785,3.052,RiskFit("Dataset 2","AIC"),RiskName("Dataset 2"))
How you view and share results varies a lot between products. Here is what each one looks like after a simulation run.
After a simulation, Analytic Solver shows thumbnail graphs for each input and output cell. Click a thumbnail to open a full-sized chart you can edit. One tip: turn off the default 3D effect, which makes the charts harder to read.
Crystal Ball opens a swarm of pop-up charts when the simulation finishes. They look reasonable, but managing dozens of floating windows and figuring out which output is which gets tiresome quickly.
ModelRisk has a separate Results Viewer that stays open during the simulation and updates live. You can set up multiple tabs with different charts and tables, organise them however you like, and export directly to various file formats. The viewer settings are saved inside the Excel file, so your graphs regenerate automatically next time you run the model.
@RISK shows a live chart during the simulation, and you can flip between outputs with ALT+right arrow. Small icons let you switch graph types. Once the simulation is done, you access results through the Reports dropdown, which takes several clicks. If you need multiple charts side by side, you end up juggling separate windows - not ideal for large models.
Decision trees map out a sequence of decisions and chance events as a branching diagram. They remain the right tool when the question is a sequence of decisions - invest or wait, test or launch - rather than a single uncertain outcome. Among the products reviewed here, Analytic Solver includes decision trees (Simulation edition) and @RISK users buy PrecisionTree as a separate product.
How well trees work in Excel depends entirely on who owns the drawing: the analyst, or the add-in. In the older tools the diagram is ordinary drawn shapes that the analyst must maintain by hand. Here is a decision tree built in Analytic Solver:
The calculations sit in cells but the connecting lines are just shapes drawn on top of the spreadsheet, and the tool does not repair them - move a row or resize a column and the lines drift out of place. The fix is not to abandon worksheet trees; it is to make the add-in own the drawing. Vose Software's ModelChoice add-in draws the tree for you, keeps it aligned, and re-solves it every time you change an assumption, so the diagram above cannot fall apart the way hand-maintained shapes do. Here is a decision tree built in ModelChoice:
Beyond the drawing, ModelChoice performs the backward-induction roll-up automatically and adds the analyses that make a tree defensible: sensitivity, risk profile, robustness, value of information (EVPI) and multi-criteria scoring. It also imports PrecisionTree models with one click, so existing @RISK/PrecisionTree users can bring their trees across. ModelChoice pairs naturally with ModelRisk: trees for the decision structure, Monte Carlo simulation for the uncertainty inside each branch. If you are weighing up decision-analysis tools specifically, we compare seven of them — including PrecisionTree, DPL, Analytica and TreeAge — on features, price and what you only find out after you buy: the top 7 decision analysis tools.
Feature testing was carried out on the versions listed below. Prices were updated in August 2026 from the sources in footnote 8.
It depends on your needs. ModelRisk offers the widest technical feature set (135 distributions, 1,400+ functions) at the lowest professional-grade price, making it the best value overall. Analytic Solver is the choice if you need stochastic optimisation today, and RiskAMP is the cheapest option for basic Monte Carlo work.
As of August 2026, published prices range from RiskAMP at $279 one-time (riskamp.com) to Analytic Solver Simulation at $2,500 per year (solver.com). Lumivero’s web shop lists @RISK at €2,900 (Professional) to €3,500 (Industrial) per year. ModelRisk costs at most €1,550 per user per year — see our price list. Oracle does not publish Crystal Ball pricing.
Crystal Ball remains available from Oracle, but it is a legacy product with a small feature set (22 distributions and around 35 functions in our test), it is not a native Excel add-in, and Capterra lists it as contact-vendor pricing with no free trial (retrieved August 2026). Most organisations evaluating it today should compare modern alternatives first — see ModelRisk vs Oracle Crystal Ball.
RiskAMP, at $279 one-time for the Professional edition or $19 per month on subscription (riskamp.com, retrieved August 2026). It covers basic Monte Carlo simulation but has a minimal feature set and limited results analytics, so it suits learning and simple models rather than professional risk analysis.
ModelRisk includes free converters for both @RISK and Crystal Ball models — the only product in this review that converts both. Analytic Solver converts @RISK models, and @RISK converts Crystal Ball models. Converted ModelRisk models remain ordinary, readable Excel workbooks.
Analytic Solver includes decision trees in its Simulation edition. PrecisionTree is no longer sold on its own — Lumivero's shop lists it only inside the DecisionTools Suite (€3,680–€4,790 per year, retrieved August 2026). Vose Software offers ModelChoice, a dedicated decision-analysis add-in with trees drawn directly on the worksheet, automatic roll-up, EVPI, robustness and multi-criteria analysis — and one-click import of PrecisionTree models. Crystal Ball and RiskAMP do not offer decision trees. We compare the decision-analysis tools in detail in the top 7 decision analysis tools.
ModelRisk Optimizer is built and will be released in the near future. Until release, Analytic Solver is the main option for stochastic optimisation in Excel — contact us about ModelRisk Optimizer timing and early access.
@RISK is a trademark of Lumivero, LLC. Crystal Ball is a trademark of Oracle Corporation. Analytic Solver is a trademark of Frontline Systems, Inc. RiskAMP is a trademark of Structured Data, LLC. Vose Software is not affiliated with these vendors. Third-party prices and listing details are quoted from the named public sources with retrieval dates shown and may change.