Monte Carlo simulation is a stress-testing method that runs roughly a thousand independent multi-year paths, each with its own randomized price, growth, and user behavior. The output is a probability distribution of outcomes, not a single forecast. It finds where a model breaks.
Judge a design by whether it survives the 5th-percentile path, not by how it looks at the median. A one-in-ten bad market is not a tail event over a multi-year horizon.
How it works
The simulation generates a large set of independent scenario paths, typically a thousand or more, each drawing randomized inputs from calibrated statistical distributions. Where a spreadsheet model gives you one number, Monte Carlo gives you a curve. That curve shows how the design performs across the full range of plausible conditions, including the tail outcomes no base-case projection would ever surface.
Each path draws from three coordinated models run together: a stochastic price process such as geometric Brownian motion, a cohort-based survival model for user growth and churn, and a Markov-chain market model that shifts behavior as conditions move between loose, normal, and tight. Running them jointly means price declines and elevated churn co-occur during stress, which is how the real world behaves.
Why it matters
Read the simulation across percentiles, never at the median alone. The median shows the expected case. The 5th-percentile path shows whether the protocol survives an adverse market. A design that passes at the median but fails at the 10th percentile carries real risk, because a one-in-ten outcome will likely arrive at some point over a four-year horizon.
We run the simulation before a capital raise for one reason: a model that has only been tested at a single scenario has not been stress-tested at all.
Example
Say a model shows a coverage ratio of 1.6 in the base case. A thousand Monte Carlo paths might reveal the ratio drops below 1.0 in 22% of paths when price declines and elevated redemptions hit at the same time. That 22% is invisible in the base case, and it is exactly the finding that changes the design.
Common mistake
Calibrating parameters optimistically, then reading the median as a forecast. Drift rates and churn assumptions borrowed from bull-market comparables produce a flattering distribution that understates real risk. Set parameters conservatively, and judge the design by the 5th-percentile outcome rather than how attractive the 50th looks.
See Tokenomics Design Services for how this applies in practice.
Know the terms but not sure how they apply to your project? That is what an engagement is for. We design, document, and stress-test the whole token economy inside the Tokenomics Data Room.
80+ projects advised. Complete tokenomics in 4 to 6 weeks.