The Monte Carlo simulator projects how your portfolio might evolve over time under thousands of randomised return scenarios. It's useful for two questions: "How long will my money last?" and "How much do I need to save?"
Running a simulation
- Open Dashboard → Simulations.
- The Scenario panel lets you set the inputs:
- Starting balance
- Annual contribution
- Expected return and volatility
- Time horizon (years)
- Whether retirement is enabled, and if so, the withdrawal years and annual withdrawal amount
- Click Prefill from portfolio if you want to use your current dashboard total as the starting balance.
- Click Run.
The simulator runs many trials with returns drawn from a distribution around your expected return, then summarises the results.
How to read the output
You'll see three things:
- Median outcome. The middle path across all trials. Half the time you do better than this; half the time worse.
- Percentile bands (10th and 90th). The shaded area shows the range of outcomes you'd see most of the time. A wide band means high volatility — the future is uncertain.
- Success probability. If retirement withdrawals are enabled, this is the percentage of trials where your balance lasted the full withdrawal period without hitting zero.
A success probability above 85% is usually considered comfortable. Below 70%, the plan probably needs adjustment — higher contributions, lower withdrawals, or a longer working life.
Where the model is weak
- Returns aren't actually normally distributed. Real markets have fatter tails, especially on the downside. Monte Carlo can understate the chance of severe crashes.
- Inflation is a single input. If inflation diverges from your assumption mid-retirement, results drift.
- Sequence-of-returns risk is captured, but average return is the dominant input. Garbage in, garbage out — if you input 10% returns, you'll see optimistic outcomes.
- It assumes you'll keep contributing or withdrawing on the schedule you set. Real life rarely cooperates.
Treat the simulator as a directional sanity check, not a prediction.