Methodology: Silver Price Research

Objective

Convert IA silver deep-research PDFs into structured institution-level target data, preserve history over time, and generate an interpretable silver target outlook.

Input Sources

• Research PDFs dropped into Data/IA_Researcher/Silver.
• Gemini extraction prompt for institution-level silver targets and scenario fields.
• Retained historical rows from silver_institution_targets_master.jsonl.

Transformation Logic

• Extract full PDF text and call Gemini to return strict JSONL rows.
• Validate schema, coerce supported numeric/date fields, and reject malformed lines.
• Deduplicate institution/publication entries and append only new rows to master history.
• Classify active numeric targets, split recent versus prior context windows, and compute KPI/dispersion metrics.
• Generate report visuals and synthesis from retained dataset (not only current-run rows).

Output Interpretation

• KPI and chart focus on extracted publication-date targets, not realised spot silver prices.
• Recent rows (latest window) drive the primary interpretation; older rows provide context.
• Rows with target_available=false are retained for coverage context but are excluded from numeric target distribution plots.

Limitations

• Coverage depends on available research PDFs and the fidelity of source documents.
• Comparability can vary across target types (point forecasts, survey averages, range midpoints).
• Extraction quality is model-dependent and can degrade on low-quality scans or ambiguous language.

Downstream Consumers

• Silver Market Analyst consumes this signal as silver_price_research for strategic (12M) context.
• Signal history and report pages use this output as a persisted silver research reference stream.