|FREQUENTLY ASKED QUESTIONS (AND ANSWERS)
The following questions and answers represent frequently asked questions by our customers. If you have other questions that may be useful to fellow users of FinanceIsland's tools,
please email us your question(s) at email@example.com.
Can I use the ROI analysis and cash flow tools to model a service business?
Yes, FinanceIsland's tools can be used to model many types of businesses, including a service business. To model revenues you need to enter number of units and net price.
In a case of a manufacturing or distribution business, units would represent the number of units produced or sold and the net price would be the price per unit. In case of a service business,
units could represent the number of customers and net price would be the value per customer. Units could also represent the number of subscriptions or licenses sold and net price would be the
price per subscription or license. Units can also represent the number of transactions with the net price being the price per transaction.
I have entered data for several time periods but I can only enter one set of simulation parameters. Why?
FinanceIsland's financial analysis tools are intended to bring simplified modeling and simulation capabilities to a broader audience. One of the simplifications is data entry of simulation
parameters. When you enter sales and expense forecasts for several time periods, the tools calculate the average data points, averaged over these time periods. These averages are then
expressed as "mean" for normal distribution, as "most likely" for uniform distribution, or as "mode" for triangular distribution. You then enter the corresponding simulation parameters
based on these average values. Although this leads only to one set of simulation parameters per line item, the simulation algorithm in FinanceIsland's tools identifies appropriate
simulation parameters for each time period. The algorithm translates the one set of simulation parameters into ratios, which are then applied to all periodic inputs during simulation.
For example, if you enter 100 units in one time period and 500 units in a second time period, the average number of units shown under simulation parameters will be 300. If you want to model
now the units to be between -30% and +40% of the entered values using the triangular distribution, for example, you would enter as simulation parameters min = 210 and max = 420. The simulation
algorithm applies the -30% and +40% to each of the periodic values. So, the units in the first time period are modeled as being between 70 and 140 units, and the units for the second time
period are modeled as being between 350 and 700 units.
This simplification in FinanceIsland's tools assumes that there is the same level of uncertainty in each time period. This is a good enough assumptions in many business cases. Should this
simplification be not appropriate for your financial model, however, we recommend developing a financial model from scratch utilizing more sophisticated Monte Carlo simulation software.
Which distribution should I use in the Monte Carlo simulation?
There are dozens of probability distributions used in Monte Carlo simulations. In the business world there are only few that represent the majority of situations. FinanceIsland offers three
distribution types in its financial analysis tools: normal, uniform, and triangular. These distributions and their parameters are described in detail in our
Monte Carlo simulation tutorial.
Many times a normal distribution is a good representation of the business situation. In order to be able to use this distribution, you need to know its mean and its standard deviation.
Although the mean is an obvious measure, standard deviation may not be an easily identifiable value. In this case, the triangular distribution may be a better choice. All the parameters
of the triangular distribution (minimum, mode, and maximum) are relatively straightforward measures and the distribution can be symmetrical or skewed to one side. The uniform distribution
is another choice, although it should be used only if all values of a metric have the same chance of occurrence.
Can I calculate standard deviation based on the inputs per time period?
No. Well, actually yes, theoretically. Theoretically, you can calculate standard deviation based on the inputs per time period, but the resulting standard deviation would be meaningless and
not usable for anything, especially not as an input for the Monte Carlo simulation with normal distribution. The input data entered per time period represents a time series, with for example
units in time period 1, units in time period 2, units in time period 3, etc. Statistically speaking, these units are different metrics, so calculating a standard deviation for different
metrics would be meaningless. Only if you had data points, for example historic data, for units in one time period, you could calculate a meaningful standard deviation for that time period.
Can I use the standard deviation of historic data as an input for the Monte Carlo simulation?
It depends. Standard deviation is a measure of spread or variability. It can be calculated based on historic data using for example Microsoft Excel's STDEV function or similar functions
found on many scientific or financial calculators. But this historic standard deviation represents the historic variability. If you expect the future variability to be the same as the
historic variability, and only then, you could use the historic standard deviation, divide it by the historic mean, and use this calculated ratio to multiply with the new mean. The resulting
number would be the new standard deviation. This new standard deviation would represent the same data variability as the old data and could be used in the Monte Carlo simulation with normal
distribution. Please make sure though that you expect the new variability to be the same as the old one.