⚙️ THIS PAGE IS STILL UNDER CONSTRUCTION
📈 Economic Time Series Forecasting Selection
ets4 is a monthly curated digest of recent working papers in economic time series forecasting.
Each issue highlights studies that bring innovative insights, models, methods, or data to the practice of forecasting — from macroeconomics to financial and energy markets.
✉️ Not subscribed yet?
Click here and join the mailing list to receive each new issue directly in your inbox.
🗓️ Latest issue — October 2025
Read it here and explore this month’s selection.
🗂️ Archive of past issues
Browse the complete list of previous editions.
View the ets4 Archive.
🧠 Methodology of ets4
ets4 combines automation and expert curation to identify and summarize the most relevant new research in economic forecasting.
Every month, the system screens hundreds of recent working papers drawn from leading academic and institutional sources — including NEP (RePEc), arXiv, and major central banks such as the ECB, Federal Reserve, and institutions such as the IMF, BIS, and CEPR.
Behind the scenes, a dedicated Python pipeline manages the entire process:
🛰️ Data Collection
The pipeline automatically retrieves and parses dozens of RSS feeds that publish newly released working papers in economics, econometrics, finance, and data science.
Each feed entry is cleaned, standardized, and time-stamped — only papers from the past 30 days are retained for evaluation.🧩 AI-Driven Evaluation
For every paper, its title and abstract are sent to a specialized large language model (LLM) that evaluates:- Whether the paper contains a clear forecasting or predictive modeling component.
- Its methodological novelty, empirical rigor, and potential practical impact.
- Its applicability to economic forecasting, even when the paper originates from another field (e.g., machine learning or energy systems).
🧮 Scoring & Classification
The model assigns a forecasting relevance score (1–10) and categorizes each paper as:- Directly Relevant → Forecasting of economic time series (macroeconomic, financial, or energy).
- Paper of Interest → Methodological innovation from another domain that could be adapted to economics.
- Not Relevant → Lacking predictive focus or purely theoretical.
Only papers with a score of 9.0 or higher are shortlisted for publication in each monthly issue.
🗂️ Curation & Publication
Shortlisted papers are formatted into two outputs:- An internal version with detailed scores for review.
- A public issue with accessible summaries and links to each paper.
These are automatically compiled into Markdown and published on this site.
In short, ets4 acts as an AI-assisted research curator — blending systematic data collection with expert rules and transparent evaluation.
The process ensures that each monthly issue highlights only the most innovative, empirically grounded, and forecast-relevant contributions emerging from the global research community.
Feedback and suggestions on ets4 are more than welcome.
🔗 You can view the full curation methodology and scoring pipeline on → GitHub.