---
title: "Monte Carlo: When?"
canonical: "https://support.55degrees.se/space/SP/4303257603/Monte%20Carlo%3A%20When%3F"
format: markdown
---
[jira cloud & data center] [azure devops] [standalone]
> *“All models are wrong, but some are useful.” - George Box*
### **When** Will It Be *Done*?
Every delivery team faces the relentless pressure to provide definitive dates. Stakeholders ask "*When will it be done*?" to coordinate releases and market commitments but most teams respond with "gut feels" or manual spreadsheets. These are often disconnected from the reality of their delivery process.
True predictability requires a strategic shift from deterministic, fixed-date estimates to probabilistic, data-driven forecasting.
### The Theory
The **Monte Carlo: When?** is a *[probabilistic](https://www.55degrees.se/blog/post/what-is-probabilistic-forecasting)* simulation chart that uses your **historical throughput** to determine the rate at which you will finish your work in the future. The chart runs *thousands* of simulations using variations in your historical data to reveal all potential outcomes and, crucially, the likelihood that you'll experience any specific outcome when you actually embark on the journey to do the set of work in question.
> ℹ️ Think of it like a lottery - your past throughput values are the numbers that go into the machine, they get mixed about and each has a chance of being picked (some more than others because some values occur more often). This process is then simulated ten thousand times to find the most possibilities and probable outcomes.
In the chart controls, users provide their own parameters - a **start date**, an **amount of items** and **forecast time unit preference** - via the Simulation Control to answer the question: “*When will will be done with this many items*?”
> ℹ️ The start date is *paramount* - like a GPS, it only gives duration *once* it has a **start date**. Then it can convert duration into an end date or a range of probable end dates with likelihoods. Like a GPS updates your arrival time as you drive, you should **rerun** the Monte Carlo: When? regularly to update the forecast as you work: **predictability is a verb** - it is something you *do* every day, not a status you reach.
> ℹ️ There’s a world of difference between saying something *will* happen or if it *might* happen. Want to learn more about how these forecasting methods differ? [https://www.55degrees.se/blog/post/probabilistic-vs-deterministic-forecasting](https://www.55degrees.se/blog/post/probabilistic-vs-deterministic-forecasting).
### The Data
The **Historical Throughput **used for the default calculation on the chart will comprise of all the *completed* work items in your data set.
We mentioned before that the Monte Carlo simulations take all of your historical throughput values to calculate a forecast. The key word here is *all*.
We *include weekends*** **(and holidays, sick days, etc) in forecasting and cycle time because flow metrics measure *total* elapsed time, not just working hours. Using calendar days offers a clearer, more predictable view of your process for these key reasons:
- Matching Customer Language
- Customers count calendar days, not your work hours or holidays. Saying an item takes 15 days means 15 calendar days; ignoring weekends causes confusion and frustration.
- Poorer Forecasts
- Although it seems precise, excluding weekends means adding days back when projecting, making forecasts longer and less accurate than including them.
- Skewing Risk Profiles
- Forecasting at the 85th percentile using working days often means planning at a higher percentile like 89th or 90th, altering your intended risk and leading to poor decisions.
- Complex Maths
- Removing non-working days complicates calculations due to varying holidays, weekends, vacations, and spreadsheet quirks causing errors.
- Native Variation
- Weekends, holidays, meetings, and illnesses naturally affect your data, so using calendar days lets forecasts reflect reality without manual adjustments.
> ℹ️ Want a full mathematical breakdown by the best? Check out Daniel Vacanti’s explanation on why weekends don’t actually inflate Cycle Time. [https://www.youtube.com/watch?v=C9DbPe-5bgQ&pp=ygUVNTVkZWdyZWVzIGRvIHdlZWtlbmRz](https://www.youtube.com/watch?v=C9DbPe-5bgQ&pp=ygUVNTVkZWdyZWVzIGRvIHdlZWtlbmRz)
There are a multitude of ways to refine the historical throughput used for the simulation - it just depends on what you’re trying to achieve.
To get you familiar with how this all works, we’ve conjured up some scenarios and best practises below.
> Macro (refined-tabs)
>
> > Macro (refined-tab)
>
> Don’t forecast apples 🍏 to oranges 🍊. If you want to know how long Epics are going to take, use the Item Filter to refine the Work Item Type to Epics. This will ensure that you’re forecasting like to like. The same goes for Initiatives or any other “container” of work. The containees such as Stories, Sub-Tasks, Bugs etc, can be forecast together - it’s mainly about not using the Parent to forecast the Child and vice-versa.
>
> 
>
> This notion can also be applied to most other filterable attributes that denote a difference in work that should be taken into account such as Team and Assignee.
>
> > ℹ️ You might be thinking “*Hold on! Don’t you forecast Epics using Stories in Portfolio Forecaster?*” and you’d be right.
> > ℹ️
> > ℹ️ We’re not trying to trick you, the difference is that the Monte Carlo simulations in Portfolio Forecaster *know* which items you currently have in progress and how many are left to start - ActionableAgile® Analytics *does not*, it’s only privy to your historical throughput.
>
> > Macro (refined-tab)
>
> A construction team 🏗️ wouldn’t use the same budget building a bungalow as they would a block of flats - context is key. If your process or project has changed, only use the historical data that’s relevant to how you work *now*, but do note that the simulation needs at least 10 data points to run a forecast.
>
> 
>
> > Macro (refined-tab)
>
> Work isn’t always released when it’s finished. It’s common practice for companies to deliver finished work in bulk releases or batching 🍪. This isn’t great for predictability as there will be a lot of days where the Throughput is 0 and then the odd day where it skyrockets.
> To combat this, use the Workflow Stages to alter your finish line from when work was *released* to when work was *finished* e.g. Release Ready, Pending Release, Hold for Launch, etc.
>
> 
>
>
> > Macro (refined-tab)
>
> Sometimes change comes as a surprise, other times it’s expected - luckily ActionableAgile® Analytics can account for *both*. As your forecast is based on your chosen historical throughput, it assumes that you’re working at the same rate or capacity as that selection - but that might not be true.
>
> There are two types of change that can affect Throughput:
>
> - **Positive**; the rate at which work is finished is expected to trend up 📈
> - A new team member, upgraded tooling, improved processes, etc.
> - **Negative**; the rate at which work is finished is expected to trend down 📉
> - Public holidays, vacations, parental leave, sickness, redundancy, deteriorated processes, etc.
>
> 
>
> If you need to factor in a positive or negative change, use the **Scale Throughput By** option in the Simulation Controls to artificially inflate or deflate the forecast.
>
> The default value is set to 1; entering a higher value indicates you have more capacity for completion, and entering a lower value indicates the opposite.
>
> 
>
> > ℹ️ Scaled Throughput By allows *decimal numbers*; if you find yourself down a team member, calculate what percentage they make up of your team and take that off of the default value.
### The Forecast
On the chart, the** Forecast **is comprised of four parts: the **Throughput Basis** and the **Throughput Date Control** manage the data going *in*, the **Histogram** and **Percentile Boxes** show the forecast coming* out*.
> ℹ️ All of these controls are toggle-able using the top right-hand options.
#### Forecast Input 📥
##### Throughput Basis
The **Throughput Basis** is located at the bottom of the chart; it is a run chart of the historical throughput data used for the forecast. Each data point represents an amount of items completed on that particular calendar day.
If you’re curious about the exact data being used, head on over to the Throughput Run Chart - the Throughput Basis is just a miniature version of that chart.
##### Throughput Data Control
The **Throughput Data Control** allows you to select a specific section of data on the Throughput Basis which will be used as the historical throughput in the the forecast. This can be useful if some of your data was generated in conditions unlike that for which your future work will be completed in (for example, you now have a different: team size, set of organisational constraints, balance of work, etc).
#### Forecast Output 📤
##### Histogram
The simulation results are shown on the **Histogram** - the centrepiece of the chart. It presents the likelihood of outcomes: so, various completion dates and how often they occurred.
This information provides us with how probable it will be to achieve any particular outcome. Hovering over each bar will tell you a specific date (horizontal axis) against the number of trials that result occurred in (vertical axis) e.g. *finished by July 24th in 86.8% of trials. *
##### Calendar
Use the **Calendar** to quickly gauge the likelihood of meeting your throughput goal - we highlight the 95th, 85th, 70th and 50th percentiles.
> ℹ️ Expressing timelines as a range (e.g., *"There is an 85% chance of completing 100 of items by July 24th"*) is transparent, data-driven, and builds lasting stakeholder trust.
> ℹ️
> ℹ️ - **50% Confidence **(*The Average*): This is a coin flip. When you're delivering a project professionally, telling a stakeholder there's a 50% chance it might fail just won't do.
> ℹ️ - **85% - 95% Confidence **(*The Professional Range*): this allows for a data-driven negotiation. When you give an 85th percentile date, you're basically saying there's some uncertainty, but you're making a promise that's likely to happen.
> ℹ️ If it seems less than likely you’ll meet your target, discuss ways to improve your chances. If you’re talking with someone skeptical about the data, ask what different steps you can take *now* for a better result. Focus on working together to change outcomes.
---
### The Controls
The Monte Carlo: When has a couple of controls that are unique to it, so we’ll be covering them first.
> Macro (refined-tabs)
>
> > Macro (refined-tab)
>
> The **Simulation Control** is where users can input the information needed to run the simulation.
>
> 
>
> - **The Start Date**
> - By default, the date is set to today as one of the best practises is to forecast how much you can get done in the time remaining to you. Make sure to re-run the forecast often to check your progress as time passes.
> - If your start date is in the past or future, you can click on the calendar icon to choose your preferred start date.
> - Picking the correct start date is crucial for your forecast. Often we misjudge when the actual start date will be. Fortunately, reforecasting often - especially as you have new information - will keep you from any late surprises!
>
> > ℹ️ When using a past date as a start date, remember that the forecast includes *all* work left to be done, as well as work already done in that timeframe. For this reason, we recommend that you avoid this approach and instead choose to forecast continually using today’s date or future dates for work that still remains.
>
> - **Until X Items Complete**
> - Simply inout the number of items you desire to complete; you can also enter a range in this field e.g. 100 - 150. This can be helpful if you’re unsure what the actual number of items will be, but you have a likely range. When you use a range, not only does the simulation use the variation of your throughput to come up with possible outcomes, it also uses a variation in the number of items you need to complete.
>
> > ℹ️ Work items often get broken down into multiple smaller stories. If you keep an eye on how many smaller items a larger one becomes, you learn something called your split rate. If you know that an item can do anything from stay the same or break down into up to 3 smaller items, you know that your workload could grow up to 30%. This means you might want to put a range of "100-130" into the simulation control.
>
> - **Scale Throughput by X**
> - If you’re interested in seeing how a potential change could affect the outcome of your forecast, use this field. It can visualise how an increase or decrease in your capacity will affect your throughput.
>
> > ℹ️ If your team of four has two members on vacation for the next sprint that you’re forecasting for, set the number to 0.5 to signify working at half capacity.
>
> - **Add Trials**
> - Situated just above the Histogram, Add Trials does exactly what it says on the tin 🥫. It allows you to increase the number of trials that the simulation runs to see if it changes your forecast - most likely not as ten thousand is statistically enough but it’s always fun to try.
>
> > Macro (refined-tab)
>
> 
>
> - Enabling the Percentiles highlights the 95th, 85th, 70th and 50th percentiles. This chart shows the likelihood of finishing the specified amount of work on each calendar day. For example, an 85% line on July 24th means "there's an 85% chance we'll finish x or more items by July 24th". So, 85% of trials met or exceeded that date.
#### All Charts
The Chart Controls listed here are for **All Charts** - that means that they will affect *all* other charts.
> ℹ️ Did you know that Jira users can save their carefully crafted data configuration as a **View**? [Here’s how!](https://support.55degrees.se/space/SP/3924886850/Views)
> Macro (refined-tabs)
>
> > Macro (refined-tab)
>
> This All Charts Control allows you to choose static dates or rolling period of time, so you can see data from a specific date range. The Throughput Basis will only show and use data from the time period chosen here.
>
> 
>
> - To get **Static Dates**, select directly on the Visible Period’s calendar or manually enter the dates you want in the designated boxes below. These dates will remain the same, hence the *static*.
> - For a **Rolling Period**, pick one of the pre-made options to the right of the Visible Period’s calendar. This period will update to the current day, hence the *rolling*.
>
> > Macro (refined-tab)
>
> This All Charts Control gives you the option to narrow down the data you’re analysing by attribute - this includes any Custom Fields brought in during the data set creation process.
>
> 
>
> For the Monte Carlo: When, this would be where you would refine the type historical throughput used in the forecast.
>
> > Macro (refined-tab)
>
> This All Charts Control designates which Workflow Stages are considered [not started], [in progress] and [finished]. Work items within each Workflow Stage will adhere to that sorting and count toward the calculations of Cycle Time, Throughput, WIP and Age.
>
> 
>
> The Workflow Stages are colour-coded so you can easily tell where the workflow starts and ends. Unchecking from the top down creates a later start, unchecking from the bottom up makes an earlier finish. Unchecking in the middle will merge your Workflow Stage with the one above - it will *not* remove or delete time spent in that Workflow Stage.
>
> 
>
> On the Monte Carlo: When, we’re looking at Throughput - which is the rate at which items cross into your last checked workflow stage. So, if you want to forecast when a specific number of items will get to a different Workflow Stage, simply uncheck them from the bottom up until the desired Workflow Stage is the first [finished] one.
#### This Chart
The Chart Controls listed here are for **This Chart** - that means that they will only affect the current chart.
> Macro (refined-tabs)
>
> > Macro (refined-tab)
>
> The Item Filter for This Chart functions in the same way as the All Charts version, the main difference is that it *only* affects the current chart you’re on and attributes you’ve filtered by in the All Charts version will be the only choices available.
>
> 
>
> > Macro (refined-tab)
>
> The Workflow Stages for This Chart function in the same way as the All Charts version, the main difference is that it *only* affects the current chart you’re on and Workflow Stages you’ve chosen in the All Charts version will be the only choices available.
>
> 