Key Takeaways
- Monte Carlo runs the model thousands of times with randomly sampled inputs, producing a distribution of possible outcomes.
- Key outputs: median IRR, 10th/90th percentile bounds, and the probability of exceeding your hurdle rate.
- Monte Carlo is most valuable for complex, large deals; simpler deals are adequately served by three-scenario analysis.
- The quality of Monte Carlo output depends entirely on the quality of input probability distributions.
Sensitivity analysis and scenario modeling test a handful of specific assumptions. Monte Carlo simulation takes a fundamentally different approach: it runs the model thousands of times with randomly sampled assumptions drawn from probability distributions, producing a distribution of possible outcomes rather than point estimates. This lesson introduces Monte Carlo concepts, demonstrates a simplified application to real estate, and explains when this technique adds value beyond traditional analysis.
How Monte Carlo Simulation Works
Instead of picking three or four scenarios, Monte Carlo simulation defines each uncertain variable as a probability distribution (normal, triangular, uniform, or custom). For example, rent growth might be modeled as a normal distribution with a mean of 2.5% and a standard deviation of 1.5%, meaning 68% of the time rent growth falls between 1.0% and 4.0%. Exit cap rate might follow a triangular distribution with a minimum of 6.0%, most likely value of 7.25%, and maximum of 9.0%. The simulation then runs the model 1,000-10,000 times, each time drawing random values from each distribution. The result is not a single IRR but a distribution of 10,000 possible IRRs, from which you can calculate the probability of achieving your target return.
Why it matters: Understanding this concept is essential for making informed investment decisions.
Interpreting Monte Carlo Output
Monte Carlo output is typically presented as a histogram or cumulative distribution function (CDF) of the target metric. From the distribution, you can extract: the median (50th percentile) IRR—the return you have a 50% chance of exceeding; the 10th percentile IRR—the return you exceed 90% of the time (a good proxy for conservative expectations); the 90th percentile IRR—the upside you achieve only 10% of the time; and the probability of achieving your hurdle rate—if your hurdle is 15% IRR, Monte Carlo might show a 72% probability of exceeding 15%. This probabilistic framing is far more informative than a single point estimate. A deal with a base case IRR of 18% might have only a 60% probability of exceeding 15%—much less attractive than it first appears.
| Percentile | IRR | Equity Multiple | Cash-on-Cash (Avg) | Interpretation |
|---|---|---|---|---|
| 5th (Worst Case) | 4.2% | 1.22x | 3.1% | Bottom 5% of outcomes — loss scenario |
| 25th (Pessimistic) | 9.8% | 1.52x | 5.8% | Below average but positive |
| 50th (Base Case) | 15.1% | 1.85x | 7.9% | Median expected outcome |
| 75th (Optimistic) | 19.8% | 2.15x | 9.4% | Above average — favorable conditions |
| 95th (Best Case) | 25.3% | 2.58x | 11.2% | Top 5% — everything goes right |
| Mean | 14.8% | 1.83x | 7.7% | Average across all 10,000 simulations |
| Std Deviation | 5.4% | 0.35x | 2.1% | Measure of outcome variability |
Monte Carlo simulation results for 20-unit apartment acquisition (10,000 iterations). Key inputs randomized: rent growth (1-5%), vacancy (3-12%), expense growth (2-5%), exit cap rate (5.5-7.5%). Source: Model output.
Why it matters: The most important output from a Monte Carlo simulation is not the average return — it is the probability of loss. In this example: - Probability of IRR < 0% (capital loss): 2.3% - Probability of IRR < 8% (below cost of capital): 18.7% - Probability of IRR > 12% (target return): 68.4% - Probability of IRR > 20% (outperformance): 22.1% Investors should set a maximum acceptable probability of loss (typically 5-10%) and reject deals that exceed this threshold regardless of the average projected return. A deal with a 25% average IRR but 20% probability of loss is riskier than a deal with 15% average IRR and 3% probability of loss.
Practical Application and Limitations
Monte Carlo simulation is most valuable for large, complex deals where the number of uncertain variables makes traditional scenario analysis insufficient—development projects, large value-add plays, and fund-level modeling. For smaller, simpler deals (stabilized 10-20 unit acquisitions), the traditional three-scenario approach provides adequate insight with far less effort. The primary limitation is "garbage in, garbage out"—if the probability distributions are poorly calibrated, the output is misleading. Monte Carlo tools include @RISK (Excel add-in), Crystal Ball, Python with NumPy/SciPy, and purpose-built platforms like Stochastic Solutions. Even without specialized tools, a simplified Monte Carlo can be built in a standard spreadsheet using random number generation and data tables.
Why it matters: Use Monte Carlo when: (1) the deal has 5+ uncertain variables with significant impact, (2) the total capital at risk exceeds $5M, (3) you need to communicate risk to institutional investors who expect probabilistic analysis, or (4) correlation between variables makes simple sensitivity analysis misleading.
Key Takeaways
- ✓Monte Carlo runs the model thousands of times with randomly sampled inputs, producing a distribution of possible outcomes.
- ✓Key outputs: median IRR, 10th/90th percentile bounds, and the probability of exceeding your hurdle rate.
- ✓Monte Carlo is most valuable for complex, large deals; simpler deals are adequately served by three-scenario analysis.
- ✓The quality of Monte Carlo output depends entirely on the quality of input probability distributions.
Sources
- NCREIF — Probabilistic Return Analysis(2025-01-15)
- Argus Software — Monte Carlo Simulation Module(2025-01-15)
Common Mistakes to Avoid
Defining input distributions without empirical basis
Consequence: Garbage in, garbage out—unrealistic distributions produce misleading probability estimates
Correction: Calibrate distribution parameters (mean, standard deviation) using historical market data for the specific property type and market
Running too few iterations for statistical reliability
Consequence: Small sample sizes produce unstable results that change significantly between runs
Correction: Run a minimum of 10,000 iterations; verify stability by checking that results converge across multiple runs
Test Your Knowledge
1.What is the fundamental concept behind Monte Carlo simulation?
2.What key output does Monte Carlo provide that scenario analysis cannot?
3.What type of probability distribution is commonly used for rent growth in Monte Carlo simulations?