Monte-Carlo Method. Monte Carlo simulation. Risk analysis applying Monte-Carlo Method
May 13, 2013
Risk analysis is an integral part of any decision we make. We constantly face uncertainty, ambiguity and variability around us. Even in spite of unprecedentedly wide access to information, we cannot precisely predict future and our project development.
Our experience in project managing and providing consultation services in Ukraine and Russia strongly shows that relatively insignificant part of companies performs risk analysis and further explanation of how the projects should be manages in the context of changes.
Most of companies, especially small, prefer act using rabbit hole logics. Do you remember, Alice in Wonderland fell into rabbit hole and has been falling, all things and events were sweeping by her and she was doing nothing, but passively observing.
Nevertheless, it is good to remember that project success constituent (and any beginning) is definitely in risk management, “it is necessary to manage risks professionally. Expert never loses in any way, even in the worst rate, and if the deal is successful, the expert takes to the maximum.” (from Preferans Guide)
Within risk management, there exist both qualitative and quantitative analytical methods. However, when it comes to “roulette risks” – all around the world may be heard words – Monte Carlo, recollecting statistical modeling method. For the first time this method was used by scientists who were designing a nuclear bomb; it was named after Monte Carlo – resort in Monaco, famous for its casinos and gambling industry.
In this article we will study Monte-Carlo method and its realization in Spider Project program complex.
Monte Carlo simulation method represents mathematical algorithm meant for risk assessment in the process of qualitative analysis and further decision making.
Within Monte Carlo method risk analysis is performed by applying possible results models. When creating such models any factor pertained to be uncertain is changed by value range, moreover, every time other random value set of probability function is used. Monte Carlo simulation permits to receive distribution of values for potential consequences.
In Spider Project the value range, necessary for calculation, is designated by direction of optimistic, expected and pessimistic parameter estimation (duration and operation volumes, labor input, productivity, quantity and resources loading, operation and resources calendars, required expenses and materials consumption) which are common to be uncertain. (see fig. 1).
Figure 1. Optimistic, expected and pessimistic parameter
Either, if there exist 3 project versions for risk analysis by three scripts method, the value may be imported from them (see fig. 2.).
Figure 2. Value import from three scripts.
Let’s view applying Monte Carlo method in the context of specifying distribution of project duration probabilities.
Main point of this method consists in performing range of simulations:
- program imitates “a throw of dice” and generates in a random way values selection, which may be possible values of duration of each task in which different values are possible;
- duration of each task is selected, critical path and also general duration and project completion date are calculated. (In Â Spider Project there exists a concept “critical index” – percentage ratio of entering the task on critical path (see fig. 3));
Figure 3. Critical index
- as a result, simulation range for each task and project there shall be specified duration and completion date which «is drawn» more often, and so its more likely value is assessed;
- as a result we get distribution of probabilities for project possible duration and completion date (see fig.4).
Figure 4. Distribution of probabilities for possible
Figure 5. Distribution of probabilities for possible
It goes without saying that using of such method shows pinpoint accuracy by more number of simulations. From time to time for modeling accomplishment, it is necessary to perform thousands and dozens of thousands of recalculations.
Method of Monte Carlo simulation is more comprehensive idea of possible events. It helps to assess not only what may happen but also the probability of such result.
Modeling by Monte Carlo Method has the range of advantages in comparison with deterministic or “point estimation” analysis.
- Probabilistic results. Results demonstrate both possible events and probability of their occurrence.
- Graphical representation of results. Nature of data received when applying Monte Carlo method permits create graphics of different consequences and probability of their occurrence. It is important when transferring the results to interested parties.
- Sensitivity analysis. With few exceptions, deterministic analysis impedes determination which of variables influences to the maximum extent the results. When Monte Carlo simulation you may see what initial data makes the most impact on ultimate results
- Script analysis. It is very complicate in deterministic models simulate various values combinations for various initial values, and, consequently, to assess influence of really dissimilar scripts. Applying Monte Carlomethod analysts may precisely define what initial data leads to certain values, and trace the occurrence of specific consequences. It is very important for performing further analysis.
This method is applied by professionals in different spheres, in particular: finance, project management, oil and gas industry, transport and environmental protection.
Monte Carlo simulation helps to study all possible consequences of your decisions and assess risk influence on project goals (terms, price, etc.), which provides higher effectiveness in decision making in conditions of ambiguity.
In the next article, there will be introduced Step-by-Step Guide to Risk Analysis in Spider Project Software Using Monte Carlo Simulation.