Forecasting Demand for New Products: Combining Subjective Rankings with Sales Data

24 Pages Posted: 18 Feb 2021

See all articles by Marat Salikhov

Marat Salikhov

New Economic School; SKOLKOVO Moscow School of Management

Nils Rudi

Yale School of Management

Date Written: February 6, 2021


A major obstacle to wider adoption of the newsvendor model is the difficulty of obtaining its key input---the demand distribution forecast, specifically when the products are new and no historical data are available. In such cases, judgmental forecasting methods are a commonly suggested solution, in particular, the Sport Obermeyer approach which collects point forecasts of demand quantity from a panel of experts and uses the degree of disagreement between experts as a proxy for demand uncertainty. However, our attempt to implement this approach at fashion retailer Moods of Norway was a failure. We were not able to recruit a sufficiently large and diverse crowd because many potential experts found it difficult to provide quantity inputs. In response to this issue, we started asking the experts to rank the products within their respective categories. While this new type of input boosted participation, its conversion to quantities requires additional data and new methodology. To that end, we propose to use category-wise historical data, and we constructed a framework for this conversion based on a tripartite decomposition of the demand vector into total demand, ordered proportions, and ranking. We also propose several new evaluation metrics and test our framework on a dataset from Moods of Norway.

Keywords: forecasting, probabilistic forecasting, wisdom of crowds, newsvendor model, subjective rankings

JEL Classification: C44, C53

Suggested Citation

Salikhov, Marat and Rudi, Nils, Forecasting Demand for New Products: Combining Subjective Rankings with Sales Data (February 6, 2021). Available at SSRN: or

Marat Salikhov (Contact Author)

New Economic School ( email )

100A Novaya Street
Moscow, Skolkovo 143026


SKOLKOVO Moscow School of Management ( email )

1st km of Skolkovo highway
Odintsovsky District
Moscow 115035

Nils Rudi

Yale School of Management ( email )

135 Prospect Street
P.O. Box 208200
New Haven, CT 06520-8200
United States

Do you have negative results from your research you’d like to share?

Paper statistics

Abstract Views
PlumX Metrics