Access Foresight

  • IIF members click here to access the current issue and archives (requires a password).
  • Free subscribers click here to access the current issue only (no access to archives).
  • To become an IIF member (full access to current issue and archives), click here.
  • To become a free subscriber (current issue only, no access to archives), click here.

 

2026:Q4 In This Issue

CSP: A Free Baseline for Probabilistic Forecasts

Valery Manokhin describes an FVA-like approach to evaluating probabilistic forecasts, called Conformal Seasonal Pools.

Commentary: FVA for Probabilistic Forecasts Stefan de Kok (who proposed the Stochastic Value Added (SVA) metric in a 2017 Foresight article), found much to praise in the CSP approach, noting that it provides an intuitive reference forecast that can be consistently applied for FVA calculations.

Conformal Prediction: Read the Fine Print!

Stephan Kolassa focuses on the conditions and assumptions that conformal prediction requires to make its guarantees, imploring us to “read the fine print.”

Forecast Explainability Is in the Eye of the Beholder

From a research survey of forecasting practitioners and interviews with business and government leaders, Nikos Kourentzes and Matthew Schneider found that “explainability” lacks a consistent understanding among different forecast users.

Seal and Score: Forecast Accountability for Pundits

Vijay Jyotish describes Seal and Score, a five-step discipline that anchors each forecast with OpenTimestamps, an open-source protocol that lets you prove a prediction existed at a specific point in time by anchoring its hash into the Bitcoin blockchain.

Maximizing Business Value with Custom Loss Functions

Janina Zittel and Milena Petkovic argue that organizations should use custom loss functions to better translate business-derived costs directly into a model’s training objective. They showcase a successful application of this approach in the natural gas sector.

An FVA Approach to Causal Demand Signals

Forecasters can purchase a wide variety of external data to enhance their models – but should they? Vishal Singh shows that the FVA framework is as useful for evaluating data as it is for its traditional use in evaluating process steps.

Response to Commentaries on “Forecasting the Impact of Artificial General Intelligence (AGI)”

Spyros Makridakis’s issue #81 essay “Forecasting the Future of Artificial General Intelligence (AGI)” drew a wide response, including six commentaries published in issue #82. Now Spyros responds to the commentators and reaffirms his call to steer AGI responsibly.

Minitutorials

  • Why Forecast Model Selection Is a Very Hard Task by Fotios Petropoulos
  • How to Best Select Forecasting Models by Fotios Petropoulos

Spotlight Interviews

  • Matthew Schneider
  • Ryan Fattini

Book Review

Ira Sohn’s reviews the book Prophecy: Prediction, Power, and the Fight for the Future, from Ancient Oracles to AI by Carissa Véliz, which provides a counterpoint to AI euphoria.