Combining Forecasts: Replication Data for the Averaging vs. Optimal Weights Puzzle

Graham Elliott's replication dataset for the paper 'Combining Forecasts - On Why Averaging Beats Optimal Linear Weights'. The 168.3 MB collection includes code and data files supporting the analysis of the forecast combination puzzle. The dataset was last updated on 2026-04 27 and is shared under a CC-BY-4.0 license.

Use Cases

• Replicating the empirical analysis of the forecast combination puzzle based on the described theoretical results.
• Testing the new weight estimation procedure introduced in the paper based on the provided code and data.
• Comparing the performance of simple averaging versus optimal linear combination weights based on the simulation or empirical data.
• Analyzing the conditions under which gains from optimal combination are small or large based on the theoretical framework.

Strengths

• Includes replication code in multiple languages (R, M, MATLAB) for full methodological transparency.
• Dataset size of 168.3 MB suggests substantial supporting material for the analysis.
• Shared under a permissive CC-BY-4.0 license allowing for reuse and modification.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

Graham Elliott

Collection Method

Likely generated for academic research and replication.

Time Range

null

Freshness

Last updated 2026-04-27 16:39:35; freshness should be verified.

Geography

null