{"slug": "historical-ai-consensus-snapshots-separating-stored-forecasts-from-evaluated", "title": "Historical AI consensus snapshots: separating stored forecasts from evaluated performance", "summary": "Future Edge Group's iPulse AI has published a versioned Hugging Face dataset of 746 rows of historical AI consensus snapshots under a CC BY 4.0 license, keeping snapshot identity, asset, forecast horizon, scoring timestamp, model/advisor counts, methodology versions and checksum together per record. The builder states the records are stored system outputs rather than a validated trading benchmark, warning that a consensus score is not a calibrated probability of being right and that advisor agreement is not independent evidence of accuracy. For reuse, the builder recommends keeping snapshot IDs and version fields in grouping keys, separating generation time from the forecast anchor, and not pooling horizons or algorithm versions into a single performance number, with any outcome study requiring a declared evaluation window, realized outcomes, a baseline, and treatment of missing or censored observations.", "body_md": "I build iPulse AI at Future Edge Group. We have published a small, versioned dataset of historical consensus snapshots for people studying how to make AI investment research inspectable.\n\nThe Hugging Face viewer currently contains 746 rows. Each record keeps the snapshot identity, asset, forecast horizon, scoring timestamp, model/advisor counts, methodology versions and checksum together. The dataset is CC BY 4.0:\n\nThe distinction that matters: these are stored system outputs, not a validated trading benchmark. A consensus score is not a calibrated probability of being right, and agreement between advisors is not independent evidence of accuracy.\n\nFor reuse, I would keep snapshot IDs and version fields in the grouping keys, separate generation time from the forecast anchor, and avoid pooling horizons or algorithm versions into one performance number. Any outcome study still needs a declared evaluation window, realized outcomes, a baseline, and a treatment of missing or censored observations.\n\nThe public product methodology is at [iPulse AI Methodology Docs | AI Market Intelligence Transparency](https://ipulseai.com/methodology).\n\nWhat metadata would you need before using a dataset like this for a leakage-safe evaluation? In particular, would you separate recorded outputs and realized outcomes into different tables, or require an immutable joined evaluation release?", "url": "https://wpnews.pro/news/historical-ai-consensus-snapshots-separating-stored-forecasts-from-evaluated", "canonical_source": "https://discuss.huggingface.co/t/historical-ai-consensus-snapshots-separating-stored-forecasts-from-evaluated-performance/182868#post_1", "published_at": "2026-10-04 09:26:46+00:00", "updated_at": "2026-10-04 09:43:02.336396+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-research", "structured-data"], "entities": ["Future Edge Group", "iPulse AI", "Hugging Face"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/historical-ai-consensus-snapshots-separating-stored-forecasts-from-evaluated", "markdown": "https://wpnews.pro/news/historical-ai-consensus-snapshots-separating-stored-forecasts-from-evaluated.md", "text": "https://wpnews.pro/news/historical-ai-consensus-snapshots-separating-stored-forecasts-from-evaluated.txt", "jsonld": "https://wpnews.pro/news/historical-ai-consensus-snapshots-separating-stored-forecasts-from-evaluated.jsonld"}}