Using Monte Carlo simulation in Risk
What is Monte Carlo Simulation?
Monte Carlo simulation is a mathematical technique that uses random sampling to estimate the probability of various outcomes in a process that’s uncertain or unpredictable. Named after the famous casino town, this technique models all possible outcomes of a process by running thousands (or even millions) of simulations. Each simulation randomly generates values for the uncertain variables within a predefined range and calculates an outcome, building a distribution of potential results. This is especially useful in risk management because it gives decision-makers insight into the likelihood of different scenarios.
Why Monte Carlo Simulation for Risk Management?
Risk is inherently unpredictable. Traditional forecasting techniques often rely on a single "best estimate," which doesn’t account for variability in complex projects or financial investments. Monte Carlo simulation, however, captures a wide range of possible outcomes and shows how each factor interacts with others to affect overall results. By presenting a probabilistic distribution of possible outcomes, Monte Carlo gives companies and project managers a nuanced view of potential risks. Here are a few ways it proves invaluable:
Quantifying Uncertainty: Instead of providing a single estimate, Monte Carlo simulation offers a range with probabilities, letting stakeholders see not just what might happen, but how likely each outcome is.
Building Resilience: Knowing the likelihood of various adverse outcomes enables organizations to prepare proactively. For example, a project manager might allocate contingency budgets based on the probability of cost overruns.
Improved Decision-Making: Monte Carlo simulations help leaders make decisions with more information. For instance, in finance, investors use it to assess risk in portfolios, seeing the probability of returns across various scenarios before committing capital.
Key steps for using Monte Carlo in Omega 365
1. Assess risk including three point estimate: When assessing risks, provide three-point estimate by specifying minimum, most likely and maximum values, in addition to the likelyhood for the risk to occour:
The formula for calculating the expected value is be default: Likelihood * (The minimum value + the most likely value * 4 + the maximum value) / 6.
The weighting values for the minimum/most likely/maximum values can be adjusted as part of the configuration of Omega 365 Risk Management.
2. Run simulation: Open the Monte Carlo app from the "Tasks" menu in the Risk Register, and select Type (cost or schedule), number of simulations / test to perform, and number of intervals for the consequence values (read more about this later in this article):
3. Review the results of the simulation: Based on the simulation, probability values, average and standard deviation is calcuated.
The meaning of the different values:
P10 (10th percentile): This represents an optimistic scenario. P10 is a cost estimate where there’s only a 10% chance that the actual cost will be lower than this amount. In other words, it’s a "low-cost estimate" that might occur under favorable conditions.
P50 (50th percentile): Also known as the "median estimate," this is the most likely cost level. Here, there’s a 50% chance that costs will be lower than the P50 estimate and a 50% chance they’ll be higher. P50 is often used as a realistic estimate, as it’s neither overly optimistic nor overly conservative.
P90 (90th percentile): This represents a conservative estimate with higher certainty. P90 is the amount where there’s a 90% chance that actual costs will be lower. It’s commonly used when a larger safety margin is needed or when the project has significant uncertainty.
Based on the number of intervals selected, it will be generated 25 value intervals, with the likelyhood to end up between these values:
In addition P-values will be generated from P1 to P100.
How is the random sampling working?
Based on the likelihood of the risk and the three-point estimate, the model repeatly selects random values for each variable, based on the triangual distribution. This means that the most likeliy outcome has a higher chance of being chosen, and values near the minimum and maximum hare chosen less frequently. For each iteration, the simulation randomly generates values within the triangular range for each variable, creating one possible "consequence" or outcome based on that iteration's values. The simulation repeats the process the number of times specificed (set prior to running the simulation). This could be millions of times.
For more information about the concept with triangular distribution, read this article: https://en.wikipedia.org/wiki/Triangular_distribution