Decision-making is one of the most common tasks we perform daily. Questions become more involved as they increase in difficulty. Should I quit my job and start my own business? How should I invest my money (assuming I™m earning more than I need)? Many of these questions can be answered numerically by probability models of some kind. Unfortunately, most people don™t have the time or resources to engineer a proper model. So we often end up forming opinions and preferences through approaches that ignore the possible underlying mechanisms behind the choice we™re trying to make. Like making an educated guess. So how can we tell the difference between good guesses and bad ones? Is there a way to quantify these guesses for decision makers to compare different alternatives more effectively and make better-informed decisions as a result? As it turns out, the answer is yes!

Four years ago, I faced a difficult choice about what to do with my life. I came up with three opportunity choices: go to school, get a job in a public organisation, or get a job in a private organisation.

I used a few frameworks, from sheer imagination to decision matrices. But my risk-averse former self would have probably preferred to use a more probabilistic approach, such as the Monte-Carlo Simulation, to measure my projected quality of life.

Decisions, Decisions¦

Let™s pretend we™re in 2018 and I want to see how a Monte-Carlo simulation can help me make this major decision.
I set up an experiment:

1. For each opportunity, I consider the five factors that are most important to my decision:

• Time Freedom
• Financial Gain
• Learning Happiness
  (Career) Progression

2. I then assign each factor™s range of values (my sentimental utility), with a maximum of 10. For example, the time freedom I will obtain from school would be somewhere between 8 and 10. I add up the Random Selection values and obtain a Quality of Life Score. The tables below show the results from 1 simulation for each opportunity presented to me.

Table 1: Quality of life score in school

Table 2: Quality of life score in Public Service

Table 3: Quality of life score in Private Sector

Every time the simulation runs, each Random Selection value changes to a random value between the minimum and the maximum.

After running 1,000 Simulations, I obtained the average or expected quality of life for each opportunity.

The Practicality of the Monte-Carlo Simulation

The appropriateness of the Monte-Carlo simulation depends on the use case. Probability-based models have applications in decisions that affect various industries from finance and technology to retail and logistics.

However, Monte-Carlo simulations are not always appropriate to use. In property valuation, for instance, inputs are often a range of values determined by industry standards. These values are adjusted by the valuer’s reasonable judgement. Although it may be an interesting exercise, it is unacceptable to provide a mean or median estimate from, say, 1,000 simulations. A similar stance is taken in many corporate finance applications, where financiers generally prefer intuitive and user-friendly models. Since Excel is pretty much the world’s choice of database, simulation model use cases in finance are few and far between.

Nonetheless, I have found Python and R to work well for simulations, especially when using large datasets. To the overachievers, you can even then feed the results of the simulations into a spreadsheet if you really must Excel!

Life is about making decisions, so it’s important to know how to deal with uncertainty. One doesn™t need to use an involved model to make easy and insignificant decisions, like when to eat. However, it may be a good idea to use a model or even a simple framework, like decision tables, to make more involved decisions. Such decisions as those that could mean the difference between a good year and tough times that last.