Evidence-Traceable Forecasting Copilot (ETFC)

Read full text
Authors: Collins Iziegbe OSOBASE
Cited by: 0

Abstract

Evidence-Traceable Forecasting Copilot (ETFC) MVP Release Author: Collins Iziegbe OSOBASE This OSF release is a Minimum Credibility Package (MCP) for an Evidence-Traceable Forecasting Copilot: a records-first approach to judgmental probabilistic forecasting for high-stakes governance decisions. The core idea is simple: forecasts are only as trustworthy as their trace. Every material claim is linked to an evidence artifact or explicitly flagged as an assumption; each forecast is stored as an audit-ready record with probabilities, rationale links, update triggers, and a contestability trail. What this release is for Creating calibrated-style probabilistic forecast records that are reviewable, contestable, and reusable. Supporting decision workflows such as triage, escalation, remediation selection, and closure acceptance. Demonstrating a concrete, auditable forecasting-to-decision infrastructure, not a general-purpose chatbot. Included in this OSF release record_builder_wizard/ A lightweight browser-based wizard for creating ETFC forecast records and exporting JSON/CSV/PDF summaries. It is records-only: no autonomous action, no retrieval, no recommendations. sample_export/ETFC-DEMO-001.json An example exported record showing the schema and linkage conventions. sample_export/ETFC-DEMO-001_summary.pdf A one-page, reviewer-friendly summary demonstrating audit-ready output. Quick start Download and unzip the release package. Option 1 (fastest): open record_builder_wizard/index.html in a modern browser and step through the record fields. Export from the Review step: JSON record CSV registers PDF summary (or use your browser Print to PDF if PDF export is unavailable) Outputs and structure Evidence register: artifacts logged with IDs (for example E001). Claims register: material claims linked to evidence IDs and assumption IDs. Assumptions register: assumptions with status and test plans. Forecast log: time-stamped probabilities with rationale claim IDs. Challenges log: contestability trail, including outcomes and any changes. Safety and scope This is downstream decision support only. It does not take actions, execute decisions, or operate autonomously. Do not include personal data or sensitive identifiers in records. Use de-identified references and governance-safe excerpts. Suggested citation Collins Iziegbe OSOBASE. Evidence-Traceable Forecasting Copilot (ETFC) MVP Release. OSF.

Fields of Study & Subjects

Forecasting Techniques and Applications Decision Support System Applications Ethics and Social Impacts of AI

Publication details

Publication date
1 January 2026
Document type
Preprint