Confidence in a forecast: what certainty does not guarantee
In Zurich, a forecast of demand, claims or sales activity can shape substantial commitments before its accuracy becomes observable. The management question is specific: how can we check whether the confidence expressed in a forecast matches its reliability?
What the literature establishes
Don A. Moore and Paul J. Healy (2008, Psychological Review) distinguish three phenomena often grouped under overconfidence: overestimating one's performance, believing oneself better than others, and assigning excessive precision to one's estimates. These phenomena do not necessarily move together and can lead to different diagnoses. For a manager committing resources on the basis of a forecast, the problem is therefore not just optimism, but also an uncertainty range that is too narrow.
A range can still convey excessive certainty
Jack B. Soll and Joshua Klayman (2004, Journal of Experimental Psychology: Learning, Memory, and Cognition) examine overconfidence in interval estimates. Their experiments show that the way an estimate's bounds are elicited can affect calibration: the correspondence between stated confidence and the frequency with which the interval contains the correct answer. Asking for a range rather than a single number is therefore not enough to ensure that uncertainty is appropriately represented.
The premature conclusion
One might conclude that ranges should always be widened or that confident people should be distrusted. This would confuse a measurable calibration problem with a personal characteristic, while overlooking how little an extremely wide range may contribute to a decision. The aim is a forecast that is both precise enough to guide action and cautious enough to represent uncertainty. (our executive and employee training programmes)
What the evidence cannot promise
These studies do not demonstrate that changing a forecasting form alone improves decisions in a bank, an insurance company or a technology business. Experimental tasks reproduce neither the internal incentives nor the market changes that affect professional forecasts. In practice, a few successful forecasts are likewise insufficient to establish good calibration, particularly when their errors depend on the same event.
A practical check in Zurich
As part of SHR's decision-making programme, a Zurich team in financial services, insurance, tech or a European headquarters can keep a register of recurring forecasts for the same observable quantity, such as monthly customer enquiry volumes. Before each deadline, the team records the forecast bounds and their stated coverage level, then retains the original forecast rather than replacing it with revisions. On a review date set in advance, it calculates the proportion of outcomes falling within the intervals and compares this with the stated coverage, while also examining the width of the ranges. This monitoring does not prove a causal improvement in decisions, but makes the gap between stated confidence and observed coverage verifiable. To go further: explore the Decision-making training in Zurich, or browse our executive and employee training programmes in Switzerland.
In pictures: Decision-making in Zurich
- decision-making
- Zurich
- research
- training zurich