Library Header Image
LSE Research Online LSE Library Services

Interpreting the skill score form of forecast performance metrics

Wheatcroft, Edward (2018) Interpreting the skill score form of forecast performance metrics. International Journal of Forecasting. ISSN 0169-2070 (In Press)

[img] Text - Accepted Version
Pending embargo until 1 January 2100.

Download (361kB) | Request a copy


Performance measures of point forecasts are commonly expressed as skill scores in which the gain in performance from using some forecasting system over another is expressed as a proportion of the gain made by forecasting that outcome perfectly. It is increasingly common to express scores of probabilistic forecasts in this form. Three criticisms of this approach are presented. Firstly, initial condition uncertainty (out of the control of the forecaster) limits the capacity to improve a probabilistic forecast and thus a ‘perfect’ score is often unattainable. Secondly, the skill score forms of the ignorance and the Brier scores are biased. Finally, it is argued that the skill score form of scoring rules destroys the useful interpretation in terms of the relative skill of two forecasting systems. Indeed, it is often misleading and useful information is lost when the skill score form is used in place of the original score.

Item Type: Article
Official URL:
Additional Information: © 2018 International Institute of Forecasters. Published by Elsevier B.V.
Divisions: Centre for Analysis of Time Series
Subjects: Q Science > QA Mathematics
Sets: Research centres and groups > Centre for the Analysis of Time Series (CATS)
Date Deposited: 10 Dec 2018 09:44
Last Modified: 19 May 2019 23:02
Funders: LSE KEI Fund, Lighthill Risk Network

Actions (login required)

View Item View Item


Downloads per month over past year

View more statistics