## Friday, October 28, 2016

### What is Monte Carlo Simulation? – RiskAMP

Summary

Despite the multitude of studies on Monte Carlo simulation that are highly sophisticated in their nature, Risk AMP a company which specializes in designing and developing statistical and stochastic models over a variety of test platforms, devised a concise article outlining directly what Monte Carlo simulations contribute to analysts and decision makers. Namely, a “Monte Carlo simulation, or probability simulation, is a technique used to understand the impact of risk and uncertainty in financial, project management, cost, and other forecasting models.”

Monte Carlo simulations are designed to assess uncertainty within a specific problem to give early warning to decision makers. In doing so, quantitative models are created to project into the future to give the analyst and decision makers the “best” estimate of the expected values or outcomes. Although on the same side of the coin, estimates generated by Monte Carlo simulations contain, “inherent uncertainty and risk, because it’s an estimate of an unknown value.” RiskAMP assess Monte Carlo simulations to be different from standard forecasting models since these simulations provide ranges of outcomes which present upper bound and lower bound limits to future possible outcomes.

The Monte Carlo simulation runs a randomized set of tasks which calculates those independently to the further calculate hundreds or thousands of randomly-selected values in the model against each other. This process of calculating outcomes is done until the model has reached the primary results in the model.

The paper discusses a basic example through job tasks over a series of months. To do this the model generates three distinct estimates to evaluate the frequency of a particular job taking place and rates of those jobs as what the probability is that they will occur frequently in the same time interval. Further, the Monte Carlo simulation, “will randomly generate values for each of the task, then calculate the total time to completion.

To the aforementioned end, it is “extremely unlikely,” that the simulation will assess that the absolute outcomes will be observed in reality to at the minimum or maximum total values delineated in the model. For instance, it is unlike something will have 100% certainty or 0% certainty. However, risk is built into the model due to the overall ability of an outcome to take place over instances which are assessed by the model to have a low probability.

Finally, like all forecasting models, RiskAMP assesses that the Monte Carlo simulation is only as good as the “assumptions” the analysts develop and build into the simulation. Additionally, they make the distinction that, “it’s important to remember that the simulation only represents probabilities and not certainty.” Regardless, RiskAMP contributes that Monte Carlo simulations can be “valuable tool[s]” when forecasting into the unknown.

Critique

As stated at the outset, this article has an effective way of portraying the complex ideas of Monte Carlo simulations with concision and without getting down in the technical mathematical jargon. That said, the article hit on a touchstone of how Monte Carlo simulations could be effective for forecasting into the future with accuracy, but did not completely flesh out those concepts. In the end, however, this article is beneficial or anyone who is looking to understand at a basic level how Monte Carlo simulations can better help develop insights and inform decision making about the future.

Source

#### 4 comments:

1. Tom, good job explaining Monte Carlo simulations and clearly summarizing the article. Did the article mention any specific situations in which a Monte Carlo simulation was successfully used?

2. Thanks, Aubrey! To answer your question, there is a section headed "For Example," where they assess their own problem of projecting how long a construction project is going to take for completion. Based on this model the Monte Carlo simulation assesses future outcomes based off of new and historical data (construction lengths) built into the model. Then it runs the data through a series of preset trials and produces percent likelihoods of a particular construction job length will take across the output trial times.

3. Tom, Nice article you found here. Clearly hits most of the highlights and limitations of the MCS. I'm wondering: did this RiskAMP company actually create their own version of a MCS or did they just explain how they work in general? If they did create their own, in what ways was it different from previous MCS? Thanks

4. Good article Tom. It is nice to get out of the super deep math for a bit to have a look at the process itself.