# Base rates: the first step in every good forecast > A base rate is how often similar things have happened before. Start there, then adjust for the evidence: the surest way to avoid overconfident forecasts. Sep 28, 2026 · Forecasting · Sikt Intelligence · https://www.siktintelligence.com/blog/base-rates-forecasting A **base rate** is how often something similar has happened before. It is the answer to the question: *out of all the cases like this one, how many turned out this way?* **Good forecasters start there, then adjust for the specifics.** Poor forecasters start with the story and never look at the numbers. It is the simplest habit in forecasting, and one of the most powerful. ## The inside view and the outside view Daniel Kahneman and Dan Lovallo described two ways to approach a forecast in a 1993 paper on bold forecasts and timid choices ([McKinsey summary](https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/how-to-take-the-outside-view)): - **The inside view** looks at this case on its own: the plan, the people, the obstacles you can imagine. It feels detailed and convincing. - **The outside view** ignores the specifics at first and asks how similar cases turned out. It feels dull, and it is usually more accurate. The inside view tends to be optimistic, because it only sees the path to success. The outside view includes all the ways similar cases actually went wrong. ## Base rate neglect People routinely ignore base rates when a vivid story is in front of them. Research by Amos Tversky and Daniel Kahneman showed that we judge likelihood by how easily examples come to mind, and by how well a case fits a stereotype, rather than by how common it is ([Tversky and Kahneman, 1974](https://www.science.org/doi/10.1126/science.185.4157.1124)). A classic example: a detailed, confident plan for a big project feels likely to finish on time. The base rate for large projects says otherwise. ## Reference class forecasting in practice **Reference class forecasting**, developed from Kahneman and Tversky's work and formalised by Bent Flyvbjerg, makes the outside view systematic ([McKinsey](https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/how-to-take-the-outside-view), [Flyvbjerg](https://arxiv.org/pdf/1302.3642)). Flyvbjerg's research on major projects found that cost and schedule forecasts are generally highly inaccurate and biased toward optimism, and that anchoring on similar past projects corrects much of that. ## How to use base rates: four steps 1. **Pick the reference class.** Which past cases are genuinely similar? "Crewed Moon landings attempted by a national space agency" is a better reference class than "big space goals". 2. **Find the base rate.** Out of those cases, how often did the outcome happen, and within what time frame? 3. **Adjust for the specifics.** Now bring in the inside view: current progress, funding, test results, politics. Move away from the base rate in proportion to how strong the evidence is. 4. **Update as evidence arrives.** New facts should shift the forecast step by step, not all at once. ## An example: space timelines Say the question is whether a space agency will meet an ambitious launch date. - **Outside view:** historically, ambitious spaceflight timelines usually slip. That base rate says: lean no. - **Inside view:** the rocket and lander are built, tests are going well, and the agency is well funded. That evidence pulls the forecast up. - **The forecast:** somewhere between the two, closer to the base rate unless the evidence is unusually strong. This is the kind of reasoning behind questions in [the Sikt feed](/#feed), such as whether China lands astronauts on the Moon before 2030. ## Why base rates cure overconfidence Starting from the base rate protects against the most common forecasting failure: being too sure. It keeps [calibration](/blog/forecast-calibration) honest, and it is also the natural baseline for a [Brier skill score](/blog/brier-score-explained): a forecaster who cannot beat "always predict the base rate" is not adding information. It is also why [AI superforecasters](/blog/what-is-an-ai-superforecaster) are built to find the base rate before reading the news. The news is the inside view. The base rate is the anchor. It is a big part of why AI forecasting improved so quickly (see [can AI predict the future?](/blog/can-ai-predict-the-future)). ## When base rates are hard to use - **Truly new situations** have no clear reference class. Use the closest analogy, and widen your uncertainty. - **Changing worlds.** A base rate from 1990 may not describe 2026. Weight recent cases more. - **Choosing the class.** Different reference classes give different answers. Try two or three and see how much they disagree. ## Key takeaways - A base rate is how often similar things have happened before. - Start with the outside view (the base rate), then adjust with the inside view (the specifics). - People neglect base rates when a vivid story is in front of them, which leads to overconfidence. - Reference class forecasting makes the outside view systematic and corrects optimistic bias. - Base rates keep forecasts calibrated and are the baseline any good forecaster must beat. ## FAQ ### What is a base rate in forecasting? How often an outcome has happened in past cases similar to the one you are forecasting. It is the starting point before you adjust for the specific evidence. ### What is the difference between the inside view and the outside view? The inside view focuses on the details of the case in front of you. The outside view looks at how similar cases turned out. The outside view is usually more accurate, and good forecasts combine both. ### What is reference class forecasting? A method that forecasts an outcome by looking at the distribution of outcomes for a set of similar past cases, the reference class, and anchoring on it. ## Sources - Kahneman, D. and Lovallo, D. (1993): Timid choices and bold forecasts: a cognitive perspective on risk taking, Management Science - Tversky, A. and Kahneman, D. (1974): [Judgment under Uncertainty: Heuristics and Biases](https://www.science.org/doi/10.1126/science.185.4157.1124), Science - Flyvbjerg, B.: [From Nobel Prize to Project Management: Getting Risks Right](https://arxiv.org/pdf/1302.3642) - McKinsey: [How to take the ‘outside view’](https://www.mckinsey.com/capabilities/strategy-and-corporate-finance/our-insights/how-to-take-the-outside-view)