Risk
Monte Carlo Stock Simulation: Reading the Probability Range
Learn how Monte Carlo simulations create thousands of possible price paths, what the percentiles mean and which assumptions can mislead investors.
A Monte Carlo simulation generates many possible future price paths by repeatedly sampling random returns from a statistical model. Instead of producing one target, it shows a distribution of outcomes.
That distribution can improve risk awareness, but it is not a crystal ball and it does not estimate intrinsic business value.
How the simulation works
A common stock-price simulation starts with historical prices, calculates periodic returns, and estimates drift and volatility. Random returns then create thousands of paths from today’s price.
For each future step, a geometric Brownian motion model commonly applies:
next price = current price × exponential function of drift, volatility and a random shock
After thousands of runs, the ending prices form a distribution. Visit the Monte Carlo model page for the model overview and run a simulation from any supported stock page, such as Microsoft.
Read percentiles correctly
The median is the middle simulated outcome: half the paths finish above it and half below. A 10th percentile outcome means roughly 10% of modeled paths ended below that level—not that there is exactly a 10% real-world probability.
Percentile bands answer a conditional question: what range appears under this model and these inputs? They do not account for everything that can happen.
Volatility widens the range
Higher volatility generally creates a wider distribution. It increases modeled upside and downside dispersion even if expected return remains unchanged.
Historical volatility can underestimate sudden regime changes. Markets exhibit jumps, changing correlations and extreme moves more often than simple normal models assume.
Drift is uncertain
Small changes in expected return compound over long horizons. Historical average return is noisy and may reflect a period that will not repeat. Using an optimistic drift can make the entire distribution look attractive without improving the company’s fundamentals.
Test multiple scenarios and pay more attention to downside ranges than to one mean outcome.
Simulation is not valuation
A price process begins with the current market price. If that price is severely overvalued, simulating returns around it does not reveal intrinsic value.
Combine simulation with the stock valuation framework. A DCF asks what the underlying cash flows may be worth; Monte Carlo shows how uncertain price outcomes may disperse.
Useful applications
- Visualizing uncertainty over different horizons.
- Comparing modeled risk across stocks.
- Stress-testing position sizes.
- Understanding how volatility changes outcome ranges.
- Challenging a single-point forecast.
Common interpretation mistakes
- Calling the average ending price a prediction.
- Treating percentile bands as guaranteed probabilities.
- Assuming historical volatility is stable.
- Ignoring dividends, dilution or structural business changes.
- Comparing simulations built with different horizons or assumptions.
- Using more simulation runs to disguise weak inputs.
Ten thousand runs reduce random sampling noise; they do not eliminate model risk. Better assumptions and honest interpretation matter more than an impressive number of paths.