ClearFactr's Simulator is a powerful tool that allows you to automatically:
When done, you can inspect individual “scenarios” to learn what kinds of inputs produce what kinds of outputs.
ClearFactr runs all of this on a computing grid in the cloud, so calculations that would take a few minutes to run locally take only seconds instead.
Start by clicking the "Run" button in the Plan window’s main toolbar. The screen will split showing your Plan on the left and the configuration window to the right. You can adjust the width of the two sides with the splitter bar in the middle.
There are two basic components to a simulation: Root cells and Monitor cells. Both of these can be (and often are) ranges of cells. The simulator works by systematically generating values for the root cells and then tracking the resulting changes of the monitor cells. In order to run a simulation, you need to tell ClearFactr which root cells to generate and which monitor attributes to track.
There are two ways you can specify these cell ranges:
Note that to configure a given simulation, you can only use one mode or the other. Which one you choose will depend on the complexity of your plan. By default, if ClearFactr sees that your plan only has one tab, the simulation panel will launch in Auto-Config mode. You can always turn that mode off (or switch modes) by clicking the Auto-Config button, shown selected, in yellow, below:
With Auto-Config on, the Roots and Monitors grids will automatically populate with the various possibilities. The upper section of the configuration window shows all of the available roots with corresponding time periods that can be manipulated for each of them. Depending on your plan, you may notice that a single plan attribute (row) can show up multiple times in this area. This will happen if the attribute contains root cells at some timepoints and formula-driven cells at others. Many attributes will likely have start and end times corresponding with the entire plan’s time period.
Again in Auto-Config mode, selecting the checkbox next to a row tells the simulator to use that row as a root attribute in the simulation. In manual mode, all the rows in the lists are used and no checkboxes are shown.
For each root you must specify:
|Start/End Time||The range of time over which you want to manipulate the attribute|
|Min||The minimum value the attribute can have|
|Max||The maximum value the attribute can have|
|Step||The size of the chunks that the min-max range will be broken into|
|Generator Type||Type of distribution of the randomly generated values|
The simulator will generate random numbers that fall between the min and max values and are multiples of the step value. The distribution of the random numbers is defined by the Generator Type.
Say your range of possibilities is 0 to 100 with a step size of 5. With a generator type of Random Uniform, values of 0, 5, 10, 15, 20, etc.., will have equal chances of being generated. With a Random Normal generator, the range of 0 to 100 will represent the edges of a “bell curve” distribution, where most of the randomly generated values will fall within a certain distance of the mean, 50. You’ll often find that a lot of aspects of your plan behave this way. For example, a rate of return might typically be around 5%, but could be as low as -15% and as high as 20%.
As you click on various root attributes, they will highlight themselves within the Plan.
The lower section of the configuration window allows to choose which attributes to track as monitors. You can chose to track them over the entire plan, or over a more specific range of time.
In this section, you can also specify values that ClearFactr will track to produce a probability graph. In the example above, we're going to track how frequently -- by quarter -- the Accumulated Net P/L was under $4 million or above $12 million. You can learn more about probability graphs here.
In manual mode, the Roots and Monitors grids start out empty, like this:
To specify cell ranges, do this:
Upon doing that, you'll see the items appear in the appropriate grid to the right, like this:
Note that in the example above, the 'Central' cell range has been clicked on, turning it green. It can then be deleted from the configuration with the 'Delete Selected Roots' button.
Lastly, you’ll need to give the simulation a name and specify how many iterations you want to run. The "Scenarios" field lets you specify the number of iterations to run, in thousands. Each iteration generates new random numbers for each of your root attributes and records new results for each monitor attribute -- so the more iterations, the more accurate your results will be. Consequentially, more iterations will also mean slower run time.
Click the Run button and sit back and watch.
As the simulation proceeds you’ll see two progress lines that will look something like this upon completion:
The green line and shaded region, charted against the right axis, shows the progress of the simulation in percent terms. It will climb steadily from 0 to 100.
The purple line represents changes in the results that the simulator has encountered. You can imagine that certain combinations of inputs will produce extreme output values. Every time the simulator encounters a new extreme, the graph will blip up. Eventually after enough iterations the results will be “stable”.
Once the simulation is complete, you can review the results in the summary graphs. Two different types of graphs are available.
The “Extremes” graph shows three lines, plotted over the chosen time period of the simulation. The lines represent the best case, worst case, and median case at each of the timepoints. Some people call this type of graph a “cone of uncertainty”.
If you click one of the points on the lines, the inputs in the plan that produced that output will be shown in the plan window to the left. The Restore button will bring the plan back to the original values.
There are actually five sets of best, worst and median values that you can inspect. “1” is the “best of the best” and “worst of the worst”, and buttons 2-5 represent additional, less extreme cases.
The Periodic tab shows two types of graphs that focus on individual timepoints within the overall simulation period.
The first graph type is a distribution of the results at any particular timepoint:
A distribution graph can be selected for any timepoint any of the monitored attributes. The results show the range of outcomes as a histogram. The x-axis will show the maximum output value for each of the “buckets” of the histogram. The y-axis shows how many iterations of the simulation (or numbers of scenarios) produced a value that fell into a given bucket.
Depending on the distributions of the input values and the formulaic relationships in your plan, you may get results that are normally distributed, or not. The important thing is that you get a sense of the extremes and also what’s most likely to happen.
If you click on the line graph icon, you’ll see a graph like this:
This view shows you fifteen individual scenario results within the simulation as a whole, for a single timepoint. There are five cases each for:
You can use the arrow buttons to step through the individual timepoints and their respective charts, or you can jump to a particular timepoint directly by selecting it from the dropdown list.