Abstract
Investment ratings are often presented as if they were timeless properties of assets. In practice, they are conditional on a forecast horizon. This study measures how often the same production research system reaches a different verdict when the horizon changes from one year to five years.
The data contain 14,400 individual configured-agent investment reports contributing to 1,247 matched one-year/five-year consensus pairs across 377 assets and four production batches. Each pair holds the asset, batch, contributing task set, underlying forecast path, and scoring procedure constant; the paired contrast changes only the point on the forecast path used for the horizon summary.
- Exact verdict changed
- 56.5%
- Direction changed
- 50.0%
- Mean annual return shift
- +6.55 pp
Five-level rating changed between matched horizons.
Negative, neutral, or positive direction changed.
Average implied annual-return estimate at five years.
Study design
- 14,400 configured-agent reports.
- 1,247 matched one-year/five-year consensus pairs.
- 377 assets across four production batches.
- The asset, batch, contributing task set, forecast path, and scoring procedure are held constant within each pair.
- The paired contrast changes only the point on the forecast path used for the horizon summary.
Directional consistency increases and configured-agent dispersion falls at the five-year horizon across the four production batches.
Source: Zenodo record 22071673 · As of 2026-08-23
Figure 2 from version 1.0 of the deposited research package.
Key findings
- Among exact verdict changes, 93.0% move upward at five years.
- Configured-agent dispersion falls at the five-year horizon.
- Directional consistency increases at the five-year horizon.
- A consensus label without its horizon can conceal economically meaningful disagreement.
Scope and limitations
Data, code, and reproducibility package
The open Zenodo research record contains the working paper PDF, accessible manuscript source, aggregate results, publication figures, auditable extraction SQL, sanitized case-study summaries, production-lineage notes, and reference audit. Cite the version-specific DOI: 10.5281/zenodo.22071673.
