fb-pixel

Contents

    What is a Monte Carlo simulation and how does it work?

    by | Nov 06, 2026 | Risk Management and Scenario Analysis

    You run a Monte Carlo simulation and see a success probability. Then the client conversation turns into a pass or fail verdict. That’s the habit worth breaking. A single number isn’t a grade. It’s a gauge of how much room the plan has when markets don’t cooperate.

    A Monte Carlo simulation repeats random sampling across uncertain inputs, such as returns and inflation, to produce a range of outcomes instead of one fixed result. Retirement planning is where that range matters most.

    Used well, it’s a client conversation framework, not a math exercise. This guide covers how the simulation works, when it earns its place over a straight-line projection, and how to explain success probability and percentile outcomes in plain terms.

    Main takeaways

    • A Monte Carlo simulation runs many trials with randomly varied inputs, producing a range of outcomes with probabilities instead of one fixed result.
    • A straight-line projection suits quick illustrations. Monte Carlo adds more for income planning, where the order of returns matters.
    • Sequence-of-returns risk shows up through path variability, which a fixed-rate projection can’t capture.
    • A success probability gauges how much room a plan has to absorb bad markets, and percentile outcomes let Advisors frame results in dollars.
    • Results depend on input assumptions and can understate extreme events, so Advisors often explain these limits and update simulations as conditions change.

    How Monte Carlo simulation works

    A Monte Carlo simulation runs hundreds or thousands of trials on uncertain inputs. Each trial randomly varies values such as investment returns or inflation. The results form a distribution of possible outcomes, weighted by how often each one showed up. That range is the whole point. It replaces a single guess with a picture of the odds.

    The method rests on repeated random sampling. The model picks a value for each uncertain variable from a set range. It runs the math and records the result. Then it starts again.

    Who invented Monte Carlo simulation

    Stanislaw Ulam and John von Neumann developed the method in the 1940s while working on the Manhattan Project at Los Alamos National Laboratory. While recovering from an illness, Ulam played solitaire and spotted a shortcut. Many random trials could estimate tricky odds faster than solving them by hand. The name comes from the Monte Carlo Casino in Monaco, a nod to the role of chance.

    The step-by-step process

    1. Identify the uncertain inputs. In a retirement plan, these usually include annual investment returns, inflation rates, and life expectancy.
    2. Assign a probability distribution to each input. The distribution sets the range of values and how likely each one is. Equity returns might centre around 6%, with a spread that reflects historical volatility.
    3. Run many trials. The model samples each distribution at random, often thousands of times, and records the result for each mix. Vanguard, for example, uses 10,000 simulations from its capital markets model to judge whether savings are enough for retirement.
    4. Combine the results. The outcomes form a distribution, often shown as a histogram or probability curve. It reveals how likely each result is across all trials.

    Where Monte Carlo simulation is used

    • Finance and risk management: Stress testing and Value-at-Risk estimates rely on simulated market paths.
    • Retirement and income planning: Advisors estimate the success probability of a plan over a 30-year horizon.
    • Project management: Teams model schedule and cost uncertainty before committing to deadlines or budgets.
    • Engineering: Reliability analysis and quality control test how systems perform under variable conditions.
    • Data science: Bayesian inference and model validation use random sampling to estimate unknowns.
    Related:  Understanding financial risk management for Advisors

    The method also has standing in Canadian risk management rules. OSFI‘s 2026 Capital Adequacy Requirements let banks base their internal market risk models on historical simulation, Monte Carlo simulation, or other analytical methods.

    Monte Carlo simulation vs. straight-line projection

    A straight-line projection applies the same fixed return to every year of the plan. A Monte Carlo simulation varies returns in each trial and shows a range of outcomes and their odds.

    The straight-line approach, sometimes called a deterministic projection, tests one condition. It checks whether the money lasts if everything goes as expected. Markets rarely move that neatly. Two clients can earn the same average return over 30 years and still finish in very different places. One retires into a bear market and draws on a shrinking base early. A straight-line projection can’t show that sequence-of-returns risk.

    The comparison below covers the differences that matter most when you choose a method for a given client conversation.

    Dimension Straight-line projection Monte Carlo simulation
    Return assumption Single fixed rate Range drawn from a probability distribution
    Output One outcome (pass or fail) Distribution of outcomes with a success probability
    Sequence-of-returns risk Not captured Captured through path variability
    Client communication “Your plan works at 5%” “Your plan succeeds in 78% of scenarios”
    Best used for Quick illustrations, early-stage conversations Stress testing, retirement income planning, plan resilience checks

    Straight-line projections still earn their place. They suit quick illustrations, early accumulation talks, and clients who just want a rough answer.

    Monte Carlo adds more as the stakes rise. Decumulation and income durability are clear cases, as is any plan where the order of returns matters as much as the average. Canada’s planning standards reflect this. FP Canada‘s 2026 Projection Assumption Guidelines use a 0.50% equity safety margin that matches a 300,000-trial Monte Carlo analysis. Most retirees need a conversation that covers what could go wrong, along with what could go right.

    See how stress testing fits your risk framework

    Stress testing methods vary by purpose and depth. This guide covers the types Canadian Advisors use most, including scenario-based and reverse stress testing.

    Read the stress testing guide

    A worked example: testing a $1M retirement plan

    Few methods stress-test a retirement income plan as practically as Monte Carlo, and a short example shows why. The average path is only one of many. (This is the basic version; our guide to Monte Carlo simulation for retirement planning covers the Canadian inputs that move the result.)

    Take a 65-year-old client with $1M in savings. They plan to withdraw $50,000 a year, indexed to inflation, over a 30-year horizon. Return assumptions can anchor to FP Canada‘s 2026 guidelines: 6.3% for Canadian equities and 3.2% for fixed income. Inflation sits at 2.1%, and the portfolio holds a balanced allocation. The simulation then runs this plan across 1,000 or more paths, each with its own sequence of annual returns.

    Some paths are kind early. Strong markets in the first few years grow the portfolio well past what the client needs. Others open with a bear market that erodes the base, and each withdrawal compounds the damage. That’s the sequence-of-returns risk a fixed-rate projection hides.

    The output is a success probability: the share of paths in which the portfolio lasts all 30 years. A result of 82%, for instance, means the money lasted in 820 of 1,000 trials.

    Assumptions shape that number heavily. Morningstar‘s 2024 retirement income study ran 1,000 Monte Carlo paths and found a 3.7% starting withdrawal rate held up at 90% success over 30 years. Long-run historical returns would have pointed to 5.5% instead. That gap is one of the reasons Advisors stress test a plan before a client commits to a spending level.

    Related:  Stress testing in risk management: what you need to know

    The simulation changes the question on the table. “Will this plan work?” becomes “How sturdy is this plan?” That shift opens room for changes, such as lowering withdrawals or delaying CPP, that a binary verdict leaves no space for. Our Stress Testing add-on runs 1,001 paths and gives Advisors a success probability and percentile ranges to anchor that conversation.

    How to read Monte Carlo results in a client conversation

    Monte Carlo output only helps once you can put it in words a client understands. That means explaining success probability, reading percentile ranges, and naming the actions the numbers support.

    What a success probability means

    A 78% success probability means the portfolio lasted the full horizon in 78% of simulated paths. In the other 22%, it ran out before the end. Clients often read that 22% as a chance of disaster. In practice, many shortfall paths miss by a small amount or run dry only in the final years.

    It often helps to frame the number as a stress-test gauge. It measures how much room the plan has to absorb bad luck. It makes no forecast about the client’s own future. For a closer look at the bands Advisors use and where the industry disagrees, see what counts as a good Monte Carlo success rate.

    How to use percentile ranges

    A histogram or probability curve groups the simulated outcomes into bands. The 10th percentile shows a poor but plausible result. The 50th is the median, and the 90th reflects a favourable path.

    Those bands let you frame the discussion in dollars. You might say, “In a tough market, your portfolio might be worth $X at age 85. In a typical market, it’s closer to $Y.” Specific figures tend to land better with clients than abstract probabilities.

    The spread between percentiles also shows how sensitive the plan is. A narrow spread points to a plan that holds up across conditions. A wide one signals heavy dependence on early returns or withdrawal levels. When the number sits below the client’s comfort level, the levers are the usual ones: withdrawals, start date, RRIF withdrawal rates, and allocation. A side-by-side scenario comparison shows the client what each one moves, and Advisor and client gain a shared language for uncertainty, sturdiness, and the changes that improve the odds.

    Turn percentile ranges into client-ready visuals

    Advisors evaluating retirement income software can see how Snap handles stress testing, scenario comparisons, and one-page plan summaries.

    Explore financial forecasting software

     

    Limitations of Monte Carlo simulation and how to manage them

    Monte Carlo output is only as good as its inputs. Weak inputs or a weak model give weak results.

    Assumption sensitivity. Off-base return, volatility, or correlation inputs give results that look sure but rest on a weak base. Volatility is how widely returns swing. Correlation is how closely assets move together. Anchoring to FP Canada’s Projection Assumption Guidelines gives you a solid starting point that clients and compliance teams know well.

    Tail risk and correlation. Tail risk is the chance of rare, extreme losses. Most simulations assume a bell curve, but real markets have fat tails. Extreme events happen more often than the model says. Correlations can also spike in a crisis, which is when mixing assets matters most. The Bank of Canada‘s 2026 Financial Stability Report notes periods of rising bond-equity correlation and high volatility.

    Related:  Cash flow analysis: how to prepare and explain concepts to clients

    Advisors can tell clients the model may understate extreme outcomes. Results guide decisions; they don’t guarantee them. Updating simulations as conditions change, and pairing them with stress cases for specific events, keeps each result a starting point for ongoing planning.

    Put Monte Carlo stress testing into practice with Snap Projections

    You’re now ready to walk a client through what a simulation shows. You can turn success probability and percentile ranges into plain terms, and you can decide when a straight-line projection is enough and when a plan calls for a full stress test.

    We built Snap Projections so Canadian Advisors can run this stress testing and bring the results into client conversations. Clients see how sturdy their plan is across likely outcomes, and you gain a clear basis for discussing changes and keeping plans current.

    See how we approach retirement stress testing and plan resilience for Canadian Advisors.

    Run 1,001 paths on your next decumulation plan

    When a client retires into a bear market, a fixed-rate projection won’t show the damage. Snap’s Stress Testing add-on gives you a success probability and percentile outcomes to anchor that conversation.

    Start a 14-day Free Trial

    FAQs about Monte Carlo simulation

    How many simulation paths do I need for the results to be reliable?

    No single number is required, but a figure in the low thousands is common practice. Results stabilize as paths increase. Rerun the same plan and check whether the success probability moves by more than a point or two. If it does, add paths. More paths smooth out run-to-run noise, but they don’t fix weak assumptions. Input quality matters more than trial count.

    What success probability should I treat as acceptable for a client?

    No universal cutoff exists. Acceptability depends on the client’s spending flexibility, other income sources, and comfort with change. A lower result may suit a client with a pension or a bequest buffer. A higher bar suits fixed essential expenses. Advisors can document the agreed threshold and revisit it at each review as a shared reference point.

    How do I know if the Monte Carlo result is overstating my client’s plan resilience?

    Check that the return, volatility, and correlation assumptions are solid. Then rerun the plan with lower expected returns or a weak early-retirement sequence. A large drop in success probability shows the plan leans on favourable assumptions. Also review the shortfall paths, not just the headline percentage. Note when the money runs out and by how much.

    What if the result is lower than my client expects?

    Treat a low result as a prompt to test specific levers. Options include withdrawal amount, retirement date, CPP and OAS timing, and spending guardrails. Change one lever at a time so the client sees what moves the number most. Confirm the inputs first, since a surprising result sometimes reflects a data entry or assumption error.

    Do I need to set up anything special before running a simulation in Snap Projections?

    A simulation runs on a completed client plan. The main prerequisite is accurate client data. Enter accounts, income, spending, and assumptions before running stress testing. Review the return and inflation assumptions first, since they drive the output. Confirm the plan covers the full time horizon and all relevant income sources before you read the results.

    What you should do now

    1. Try Snap Projections free for 14 days.
    2. Read more articles in our blog.
    3. If you know someone who’d enjoy this article, share it with them via Facebook, Twitter, LinkedIn, or email.
    The Financial Advisors Marketing Guide.

    NEW

    The Financial Advisor’s Marketing Guide

    Learn how to build an online presence, create and share valuable content, and engage with prospects and clients through email marketing.