top of page

Refining Project Prediction: Reference Class Forecasting

Exploring Reference Class Forecasting (RCF)

Reference class forecasting is a probabilistic forecasting tool to inform risk and affordability. RCF also provides insights into prioritisation of projects and resource allocations. The RCF method makes use of historical data based on a reference class of past similar projects. These similar projects are determined based on statistical similarity with the project being assessed. This is designed to deal with the optimism bias that plagues project capital cost estimations.

From a RCF perspective, the main causes of mega project risks are internal as initial underestimations are the reason for later overruns. RCF adopts an outside view which eliminates the strategic misrepresentation and psychological biases that comes with inside view estimations.

Crafting Better Estimates: The Steps of RCF

First step is to build the reference class by selecting past similar projects based on statistical similarity. More data points result in a more robust reference class, particularly because RCF is a probabilistic method. Leveraging the full distribution of historical data enhances the accuracy of estimate. With First of a Kind (FoaK) project types, where there are few projects of a similar scope, data points should be other FoaK projects. Since RCF is about the similarity of risk distribution, use proxies for similar complexity, schedules and budget.

The next step is to establish a probability distribution for the reference class.

  1. Regress the best guess from the inside view towards the mean (most likely case) of the reference class. The inside view expert’s forecast will typically see the use of QRA (Quantitative Risk Assessment).

  2. Expand estimates to the full interval of the reference class which with wider intervals due to bigger variations as RCF includes everything that happens in previously completed projects. It will also be positively skewed as projects are more prone to overruns than underruns, and overruns are typically larger than underrun. Finally, your project can be compared with the distribution and translated into a probability.

Common Misconceptions of RCF

As RCF uses data from past similar projects, one misconception is history repeating meaning that we do not want to perform as badly as the projects who did not do well. There is the fear that if we use the forecasts of the lower performing projects, our project will also do badly. This is an understandable thought however, RCF reveals the data of these types of projects and other similar projects to helps us be aware of why the projects performed the way they did so we can learn from them. From this data, we can effectively mitigate the risks and outperform the previous projects.

Another misconception is what similarities to look for when wanting to use RCF. RCF is not about the exact similarity of technology, but rather about the similarities in the risk distributions between the projects. By analysing the risk distribution of previous projects, we can investigate their budgets and schedule to see how they affected the project performance. We can overall see what sort of risks occurred in the projects so we can strength our mitigation procedures for our own projects.

RCF recommends some uplifts such as inflating our project cost and schedule time. Many of us would think these numbers would look scary to key stakeholders and indeed, high-cost numbers and potential long schedule times do not look appealing. There is a fear of our project being never approved. When feeling worried about these uplifts, remember that RCF should not be used as a standalone tool. RCF needs to be used in conjunction with other tools to ensure we are enhancing the accuracy of our predictions.

Beyond Biases: Megaproject Forecasting Reality

Dr Michala Techau from Oxford Global Projects mentions that there are three main factors which could contribute to risk underestimation; technical, psychological, and political-economic biases.

Project managers need to be aware that technical and behavioural biases could influence the decision-making process in project forecasting.

An interesting clarification about economic-political bias as a factor contributing to Bent Flyvberg's Iron Law was that strategic misrepresentation is often a result of tacit culture and not from someone who wants to see the downfall of the project. Many government-related megaprojects are susceptible to strategic misrepresentation, a deliberate distortion of project costs, timelines, or benefits, rather than a mere miscalculation. This intentional skewing can stem from various factors, such as political pressures, people approvals, or personal gains, underlining the complex motivations that drive decision-makers to present overly optimistic project forecasts.

By integrating RCF, policymakers and project managers can foster more efficient project delivery through enhanced incentives. This approach also brings about a more objective foundation for decision-making, ensuring resources are distributed more efficiently. Crucially, RCF tackles the prevalent "selection effect" dilemma—where less viable projects are chosen over better ones due to superficially attractive proposals that underestimate costs and overestimate benefits.

"RCF has fundamentally changed my perspective on project management, showing me the importance of data-driven decision-making and the pitfalls of over-optimism in project planning."

Watch the video by the Association for Project Management to learn more about RCF:


Unlock the power of RCF for smarter project forecasting and management! Connect with us today to learn how historical data can drive project success. Click below to discover more and start maximizing your project outcomes!

20 views0 comments

Recent Posts

See All


bottom of page