Tuesday, November 13, 2018

Summary of Findings: Monte Carlo Simulation (4 out of 5 Stars)


Note: This post represents the synthesis of the thoughts, procedures and experiences of others as represented in the articles read in advance (see previous posts) and the discussion among the students and instructor during the Advanced Analytic Techniques class at Mercyhurst University in November 2018 regarding Monte Carlo Simulation as an Analytic Technique, specifically. This technique was evaluated based on its overall validity, simplicity, flexibility and its ability to effectively use unstructured data.

Description:
Monte Carlo Simulation is arguably both a method and a modifier by which analysts can pull together pieces of information in the form of ranges and, through random sampling of those ranges, produce an estimate. It is a highly flexible method due to its ability to handle uncertain data from virtually any discipline. The mental modeling required to piece apart the given problem decreases the simplicity of the method, but plugging in the ranges themselves is simple indeed. Monte Carlo Simulations have the capacity to produce a highly accurate distribution of estimates.

Strengths:
  1. Applicable with simple ranges (exact numbers not required)
  2. Increases confidence in estimate through statistical analysis
  3. Simulations produce a visual statistical distribution  
  4. Allows analyst to identify collection needs to improve estimate
  5. Allows analyst to create an estimate with uncertain data

Weaknesses:  
  1. Qualitative estimates require coding
  2. Statistical data can be misleading
  3. The result of a MC simulation is a numeric range, which is almost never where the analysis itself will stop
  4. Challenging to explain to decision-makers

How-To:
  1. Identify the variables that are needed to answer the question (See Application of Technique)
    1. Can be ranges or exact numbers
  1. Enter the variables into a software program capable of running a Monte Carlo simulation (Guesstimate, Microsoft Excel, @RISK, etc.)
  2. Multiply average outputs of simulations together to produce estimate
  3. Continue adding variables and running simulations until question is answered

Application of Technique: 


Utilizing www.getguesstimate.com we presented the group with a question, “How many Big Mac’s will McDonald’s sell at locations in Erie in a week?”.  The group then created a model to determine the number of Big Mac’s sold.  The group developed a series of questions with estimate ranges to answer each question.  Pictured below is the model the group created to work through the question.  The arrows connecting boxes within the model indicate that a formula was used in order to produce an answer in the form of a range of possible outcomes (i.e. 4.9 McDonald’s in Erie * 400 customers per day = 2000 people visiting McDonald’s in Erie per day)


For Further Information:
  1. Get Guesstimate


Monday, November 12, 2018

Monte Carlo Method in risk analysis for investment projects


By Victor Platona and Andreea Constantinescua

Summary:
Investment projects worth over € 50 million that are financed with EU support must meet certain conditions, including criteria related to the size of the risk. For major projects, risk analysis indicates whether risks have been taken into account in estimating the costs. In this study, the authors propose a risk estimation method, Monte Carlo method, to be applied in a standardized way on these investment projects. In the Monte Carlo method, artificial values of a probabilistic variable are generated through a random uniformly distributed number generator in [0, 1] intervals. The method algorithm is shown in its succession interactive five steps:

Step 1: Creating a parametric model, y = f(x1, x2, ..., xq);
Step 2: Generation of random input set of data, xi1, xi2, ..., xiq;
Step 3: Effective calculations and memorizing results as yi;
Step 4: Repeating steps 2 and 3 for i = 1 to n (n 5000);
Step 5: Analyzing the results using histograms, confidence intervals, other statistic indicators resulting from the simulation, etc.

The authors used 23 waste management projects and a number of 40 water and wastewater projects, which have been contracted and are under implementation. They then calculated average, standard deviation and relative standard deviation for the two types of projects.


The water supply projects had larger average value of a project is €106.23 million compared to waste management projects average of € 31.7 million. They then estimated the risk of exceeding the project value.

This showed that the average is € 50.75 million, slightly lower than the chosen project analysis, € 51.76 million. The asymmetry is 0.05 - which shows a slightly expanded distribution to the right and the flattening is -0.1 - indicating a slight flattening compared to normal distribution. The authors also estimated the risk of exceeding the project implementation period. Some of their over all conclusions included that competitive bidding system can make the offering price lower than the starting price. There are higher chances of exceeding the initially established period due to multiple situations occurring during implementation. Finally that Monte Carlo method is relatively easy to perform and provides important information regarding the risks of investment projects.

Critique:
The authors laid out a step-by-step way evaluate both the risk of exceeding the project value and the risk of exceeding the project implementation period. If the Monte Carlo method is applied correctly and with accurate inputs it can be useful in evaluating the risks in investments. Like many methods bad data will cause miss leading results that could be very costly in the financial industry.

Link: https://core.ac.uk/download/pdf/82113269.pdf