Reference Class Forecasting Calculator
The forecasting method that corrects optimism bias by pricing your project against how similar projects actually turned out. Apply the official Green Book uplift with the calculator, then read the five-step method underneath.
What is reference class forecasting?
Reference class forecasting (RCF) is a method for predicting a project's cost or schedule by looking at the actual outcomes of a large class of similar past projects, rather than building a bottom-up estimate that assumes everything goes to plan. You take an "outside view": identify a reference class of comparable completed projects, build the distribution of how far they overran, pick a confidence level, and apply the matching uplift to your estimate. Developed from Kahneman and Tversky's work on optimism bias and operationalised for projects by Bent Flyvbjerg, it is the only forecasting method with documented evidence of reducing megaproject overruns, and the UK Treasury and Department for Transport have required it since 2003. Use the calculator below to apply the official HM Treasury Green Book uplift to your own estimate.
Reference Class Forecasting calculator
Uplifts your base estimate using the empirical overrun for its reference class, the method Flyvbjerg and the UK Treasury recommend to correct optimism bias.
New build, standard design, greenfield. Flyvbjerg 'Buildings' mean: +62%.
The Green Book applies the full upper-bound uplift at the outline-business-case stage, then reduces it as the contributory risk factors are actively identified and mitigated. Set this to how much of that risk work you have genuinely done, not how confident you feel.
Uplift percentages: HM Treasury Supplementary Green Book Guidance on Optimism Bias (Mott MacDonald, 2002), Table 1, capital expenditure upper and lower bounds. Cross-check base rates: Flyvbjerg & Gardner, How Big Things Get Done (2023).
HM Treasury Green Book optimism-bias uplifts by project type
These are the official UK figures the calculator applies. The upper bound is the average historic optimism bias at outline-business-case stage; you apply it in full at the start, then reduce toward the lower bound as you actively identify and mitigate the contributory risk factors. All figures are for capital expenditure unless noted.
| Project type | Upper bound (unmitigated) | Lower bound (fully mitigated) |
|---|---|---|
| Standard buildings | 24% | 2% |
| Non-standard buildings | 51% | 4% |
| Standard civil engineering | 44% | 3% |
| Non-standard civil engineering | 66% | 6% |
| Equipment / development (incl. IT & software) | 200% | 10% |
| Outsourcing | 41% | 0% |
Source: HM Treasury, Supplementary Green Book Guidance: Optimism Bias (Mott MacDonald, 2002), Table 1, capital expenditure. The outsourcing row is measured on operating expenditure rather than capital, so treat it separately. Equipment/development covers ICT and software, which is why it carries the largest uplift. Verified against the primary guidance September 2026.
Uplifts by confidence percentile: the Department for Transport table
The Green Book guidance above gives a single upper and lower bound per project type. It does not attach percentiles to them, and it sets no default confidence level. The percentile figures come from a separate document: the Department for Transport's 2004 optimism-bias guidance, prepared by Bent Flyvbjerg in association with COWI, which built cost-overrun distributions from completed UK and international transport projects and read the required uplift off each distribution at five confidence levels.
This is the table to use when you have decided how much overrun risk the sponsor will carry. Reading across a row: an 80th-percentile uplift is the one that leaves roughly a 20% chance of overrun. Roads covers motorway, trunk and local roads plus bus lane schemes, park and ride, bicycle and pedestrian facilities and guided buses on wheels; Rail covers metro, light rail, conventional rail, high speed rail and guided buses on tracks; Fixed links covers bridges and tunnels.
| Category | 50th | 60th | 70th | 80th | 90th |
|---|---|---|---|---|---|
| Roads | 15% | 24% | 27% | 32% | 45% |
| Rail | 40% | 45% | 51% | 57% | 68% |
| Fixed links | 23% | 26% | 34% | 55% | 83% |
Source: The British Department for Transport, Procedures for Dealing with Optimism Bias in Transport Planning: Guidance Document (June 2004), Table 6, applicable capital expenditure uplifts for selected percentiles, constant prices. Distributions drawn from 172 road projects and 46 rail projects. Building, IT and civil engineering projects are not in this table: the guidance carries them only as the Green Book / Mott MacDonald ranges above, with no probability distribution available. Verified against the primary guidance September 2026.
The five steps
- Identify the reference class. Find past projects that are genuinely comparable in scope, technology, scale, and delivery model. Not "similar in name", similar in structure.
- Establish the distribution of outcomes. For each reference project, calculate the cost overrun against its original approved budget (not against any later re-baseline). Build a distribution.
- Compare your project to the distribution. Identify any structural reasons your project should sit lower (proven technology, smaller scope) or higher (less mature supply chain, first-of-a-kind) than the reference class.
- Pick a target percentile. A 50th-percentile uplift accepts a 50/50 chance of overrun. An 80th-percentile uplift gives roughly 80% confidence of staying within budget. The Department for Transport guidance is explicit about the choice: the 50% percentile “should be used only in instances where investors are willing to take a high degree of risk that cost overrun will occur” or where underruns on one project can cover overruns on another, while “the upper percentiles (80-90%) should be used when investors want a high degree of certainty that cost overrun will not occur, for instance in stand-alone projects with no access to additional funds beyond the approved budget.” The percentile is the sponsor's risk decision, not a fixed default.
- Apply the uplift. Add the percentile-derived uplift to the bottom-up budget. Document the assumptions; the uplift is a forecasting tool, not a slush fund.
Worked example: a UK light-rail project
Suppose you are forecasting the cost of a 12-mile urban light-rail extension. Bottom-up estimate is 800M GBP.
Light rail sits in the Department for Transport's Rail reference class, alongside metro, guided buses on tracks, conventional rail and high speed rail. Its distribution was built from 46 completed rail projects. The uplift column below is published in the DfT guidance; the budget column is that uplift applied to the 800M base estimate, so you can redo the arithmetic yourself.
| Percentile | Published uplift (rail, capex) | 800M base becomes |
|---|---|---|
| 50th (median) | +40% | 1.40x = 1,120M |
| 60th | +45% | 1.45x = 1,160M |
| 70th | +51% | 1.51x = 1,208M |
| 80th | +57% | 1.57x = 1,256M |
| 90th | +68% | 1.68x = 1,344M |
At the 80th percentile, the level the DfT guidance points stand-alone projects toward, the RCF-adjusted budget is roughly 1.26bn GBP, not 800M. A sponsor who publishes the 800M figure is choosing to accept the 50/50 position or worse: on the same distribution, 800M is below even the median-adjusted 1.12bn.
Uplifts: The British Department for Transport, Procedures for Dealing with Optimism Bias in Transport Planning: Guidance Document (June 2004), Table 6, applicable capital expenditure uplifts for selected percentiles, constant prices. Prepared by Bent Flyvbjerg in association with COWI. Verified against the primary guidance September 2026.
Reference-class base rates by project type
Step 2 of the method needs a distribution of real outcomes. If you cannot build your own reference class from scratch, start from the published base rates in Flyvbjerg and Gardner's How Big Things Get Done (2023), drawn from a database of more than 16,000 projects across 20-plus categories. Read the mean overrun for your category, then note the fat tail: the share of projects that blew past 50% over budget, and how far that group overran on average. The fat tail, not the mean, is what reference class forecasting is designed to price in.
| Project category | Mean cost overrun | % in the fat tail (50%+ over) | Mean overrun of that tail |
|---|---|---|---|
| Solar power | +1% | 2% | +50% |
| Energy transmission | +8% | 4% | +166% |
| Wind power | +13% | 7% | +97% |
| Tunnels | +37% | 28% | +103% |
| Rail | +39% | 28% | +116% |
| Airports | +39% | 43% | +88% |
| Buildings | +62% | 39% | +206% |
| IT | +73% | 18% | +447% |
| Nuclear power | +120% | 55% | +204% |
| Olympic Games | +157% | 76% | +200% |
| Nuclear waste storage | +238% | 48% | +427% |
Source: base-rate appendix of Flyvbjerg and Gardner (2023), How Big Things Get Done, Penguin Random House. Overruns are real (inflation-adjusted) against the budget at the decision to build. Full category breakdown and methodology: the Flyvbjerg megaproject database.
Frequently asked questions
What is reference class forecasting?
Reference class forecasting (RCF) corrects optimism bias by basing a project's budget on the actual outcomes of a class of similar completed projects, rather than on a bottom-up estimate that assumes everything goes to plan. You identify a reference class of comparable projects, build the distribution of their cost overruns, pick a confidence percentile, and apply the corresponding uplift. It is the only forecasting method with documented evidence of reducing megaproject overruns, and the UK Treasury and Department for Transport have required it since 2003.
What uplift should I apply for optimism bias?
The HM Treasury Supplementary Green Book Guidance on Optimism Bias (Mott MacDonald, 2002) publishes upper-bound capital-expenditure uplifts by project type: standard buildings 24%, non-standard buildings 51%, standard civil engineering 44%, non-standard civil engineering 66%, and equipment/development including IT and software 200%. These upper bounds apply at outline-business-case stage and reduce toward the lower bound (2% to 10% depending on type) as project-specific risk is identified and mitigated. The calculator above applies these figures directly.
Which confidence percentile should I use for an optimism-bias uplift?
There is no fixed default. The HM Treasury Supplementary Green Book Guidance on Optimism Bias publishes a single upper and lower bound per project type and attaches no percentile to them. The percentile choice comes from the Department for Transport's 2004 guidance, which says the 50% percentile should be used only where investors are willing to take a high degree of risk that cost overrun will occur, or where underruns on one project can cover overruns on another, and that the upper percentiles (80 to 90%) should be used where investors want a high degree of certainty that cost overrun will not occur, for instance in stand-alone projects with no access to funds beyond the approved budget. The DfT's published capital-expenditure uplifts run: roads 15% at the 50th percentile rising to 45% at the 90th; rail 40% to 68%; fixed links 23% to 83%.
Is there a free reference class forecasting calculator?
Yes, the calculator at the top of this page. It applies the Green Book optimism-bias uplift for your selected project type to your base estimate, adjusted by how much project-specific risk you have mitigated, and cross-checks the result against the Flyvbjerg & Gardner (2023) mean overrun for the nearest reference class. It is free, requires no signup, and runs entirely in your browser.
Sources
- HM Treasury Green Book, supplementary guidance on optimism bias
- The British Department for Transport (2004). Procedures for Dealing with Optimism Bias in Transport Planning: Guidance Document, June 2004. Prepared by Bent Flyvbjerg in association with COWI. Table 6, applicable capital expenditure uplifts for selected percentiles.
- Flyvbjerg B. (2008). Curbing optimism bias and strategic misrepresentation in planning: reference class forecasting in practice. European Planning Studies 16(1).
- Kahneman D., Tversky A. (1979). Intuitive prediction: biases and corrective procedures. Management Science 12.