Forecasting Methods and Principles: Evidence-Based Checklists

Abstract

Problem: Most forecasting practitioners are unaware of discoveries from experimental research over the past half-century that can be used to reduce errors, often by more than half. The objective of this paper is to improve forecasting accuracy by providing evidence-based methods and principles to forecasters and decision-makers in a form that is easy for them to understand and use.

Methods: This paper reviews research on forecasting to identify methods that are useful for forecasting, and those that are not, and to develop checklists of evidence-based forecasting to help forecasters and their clients. The primary criterion for evaluating whether or not a method is useful is predictive validity as assessed by evidence on its contribution to ex ante predictive validity.

Findings: This paper identifies and describes 17 evidence-based forecasting methods and 8 that are not. Six checklists are provided to apply evidence-based findings to forecasting problems by guiding the selection of the most suitable methods and their implementation.

Originality: Four of the six checklists are new: They address (1) evidence-based methods, (2) regression analysis, (3) uncertainty, and (4) popular but not validated forecasting methods. Another checklist, the Golden Rule, was improved.

Usefulness: The checklists are low-cost tools that forecasters can use to obtain forecasts that are substantially more accurate than those provided by commonly used methods. The completed checklists provide assurance to clients and other interested parties that the resulting forecasts were derived using evidence-based procedures.

Keywords: big data, combining forecasts, decision-making, decomposition, equal weights, expectations, extrapolation, index method, intentions, Occam’s razor, prediction intervals, regression analysis, scenarios, uncertainty