future-edge-group/ipulse-ai-batch5-advisor-forecast-panel

iPulse AI Batch 5 Advisor Forecast Panel

This dataset exposes a compact, anonymized panel of production forecasts from iPulse AI, Future Edge Group's Open Agentic Investment Research Platform. It is designed for research on forecast combination, disagreement, correlated errors, regime dependence, and the effective number of independent forecasters.

The release contains seven showcase assets, twelve advisor configurations per asset, quarterly forecast paths extending five years, 1-year and 5-year horizon summaries, structured drivers and risks, and realized close-price paths for ongoing evaluation.

Why this panel is useful

The twelve voices are deliberately not described as twelve independent models. They are twelve advisor configurations constructed from:

• one shared model family;
• one frozen input snapshot per asset;
• eight underlying analyst/persona identities;
• different investment-framework and reasoning-mode combinations;
• researcher modes with web search and thinker modes without web search.

This dependence is part of the research question. Low dispersion can reflect shared inputs or model bias, while a contrarian or specialist configuration can be useful precisely when the majority is wrong.

Coverage

Item	Coverage
Production batch	5
Forecast cohort date	2026-06-05
Assets	7
Advisor configurations per asset	12
Underlying analyst/persona identities	8
Model families	1
Forecast steps per asset/advisor	20 quarterly steps
Forecast-path rows	1,680
Advisor/horizon summary rows	168
Structured driver/risk rows	1,008
Realized price-path rows	465
Realized observation cutoff	2026-08-24

Assets:

• Accenture (ACN.NYSE)
• Adobe (ADBE.NASDAQ)
• Avalanche (AVAX-USD.CC)
• Ethereum (ETH-USD.CC)
• Northrop Grumman (NOC.NYSE)
• Shell (SHEL.LSE)
• Solana (SOL-USD.CC)

Dataset configurations

forecast_paths

One row per asset, advisor configuration, and quarterly forecast step. Step returns must be compounded, not summed. The file includes the resulting cumulative forecast return and implied price path.

horizon_summaries

One row per asset, advisor configuration, and 1-year or 5-year horizon. The path-based return is taken from the compounded quarterly path. A separately reported model-level 5-year base return and rating are preserved where available because they are a distinct source output and can differ from the compounded path.

drivers_and_risks

Structured model-generated horizon drivers, frictions, tail opportunities, and tail risks. These labels are research outputs, not verified events or recommendations.

advisor_configurations

Public configuration descriptions keyed by anonymous codes A01 through A12. Raw prompts, internal identifiers, provider payloads, and unrestricted model responses are excluded.

realized_prices

One row per asset and observed close date from the forecast anchor through the extraction cutoff. The file includes the actual anchor price, observed close, price-only realized return, and a realized index beginning at 100. Equity observations use trading days; crypto observations use daily closes.

Important interpretation

• This is a small, intentionally selected showcase panel, not a representative market sample.
• Neither the 1-year nor 5-year forecast horizon has resolved; the realized series is therefore partial and intended for ongoing evaluation.
• Forecasts and drivers are model-generated and may be wrong, inconsistent, or poorly calibrated.
• Forecast timing used a frozen input snapshot within each asset. Most requests were launched within minutes, but the release should not be interpreted as proof that every provider call completed simultaneously.
• The files support methodological research; they do not establish investment performance.
• Nothing here is personal investment advice or an instruction to buy or sell.

Privacy and exclusions

The release excludes raw prompts, full generated theses, unrestricted model responses, input-document payloads, storage paths, provider request/response payloads, and internal task, analyst, persona, snapshot, and configuration identifiers.

Version 1.1.0 includes actual anchor prices, absolute implied forecast prices, and realized close-price paths. Normalized forecast and realized indexes beginning at 100 remain available for cross-asset comparison.

Reproducibility

The released transformations are documented in the methodology. Run the standard-library validator:

python3 scripts/validate_dataset.py

See the data dictionary and quality report.

License and citation

The dataset and public documentation are licensed under CC BY 4.0. Attribution should identify iPulse AI and Future Edge Group FZE and cite dataset version ipulse-batch5-advisor-panel-v1.1.0.

Suggested citation:

Future Edge Group FZE. “iPulse AI Batch 5 Advisor Forecast Panel.” Dataset version 1.1.0, forecast cohort dated 2026-06-05.

Contact

Product: iPulse AI
Organization: Future Edge Group FZE
Founder: Russlan Ramdowar