Interactive timeline showing:

• phase bands (Setup, Stress, Trigger, Cascade),
• current and historical cycle classification,
• highlighted collapse windows (including an estimated 1927-1933 segment).
• rolling volatility chart for both stress score and phase regime changes (12-month window).
• cause/effect evidence explorer that auto-indexes papers and articles per data point with deterministic summaries.

python3 tools/build_timeline_data.py
python3 tools/build_sector_data.py
python3 tools/build_evidence_supplemental.py

This writes:

• data/timeline_data.json from live FRED series
• data/sector_data.json and data/combined_data.json for sector and combined dashboards
• data/evidence_supplemental.json for supplemental Fed factor references

Serve the folder and open index.html:

python3 -m http.server 8000

Then visit:

http://localhost:8000/index.html

Additional pages:

• http://localhost:8000/sectors.html (sector proxy dashboard)
• http://localhost:8000/combined.html (macro + sector combined view)

This repo includes .github/workflows/deploy-pages.yml to publish the dashboard.

In GitHub:

1. Open Settings -> Pages
2. Under Build and deployment, set Source to GitHub Actions

• On every push to main, GitHub Actions will:
  1. run python3 tools/build_timeline_data.py
  2. run python3 tools/build_sector_data.py
  3. run python3 tools/build_evidence_supplemental.py
  4. package index.html + sectors.html + combined.html + data/
  5. deploy to GitHub Pages

For this repository name, the site URL format is:

https://johnfiron.github.io/johnfiron/

• The Great Depression window is intentionally labeled estimated because high-frequency modern market stress inputs are not available for that period.
• Hover any point to see the date, phase, stress score, and core inputs.
• The dashboard now embeds a data request status panel showing every live source call, row counts, date coverage, and any fetch errors, so missing series are transparent and traceable.
• The evidence explorer uses click-triggered live search for papers/articles from OpenAlex and Crossref, based on the clicked datapoint context (metric, value, phase, stress score, date).
• Evidence retrieval now scans larger live result sets (OpenAlex + Crossref), then performs deterministic extraction per reference:
  • extracted date mention (month/year where available),
  • extracted numeric/value mention,
  • extracted causal sentence (when present),
  • deterministic sentiment label/score.
• Evidence is cross-referenced across sources for corroboration signatures (shared date/value/causal terms/sentiment) and exposes consensus support counts per item.
• References are bucketed into Prior observation, Current information (as of point), and Future analysis on all dashboards.
• Supplemental Fed report factor references are prebuilt in data/evidence_supplemental.json via tools/build_evidence_supplemental.py.
• Phase is calculated (not manual) from weighted component scores with explicit phase transition rules; these rules are shown in the UI under How phase is determined.
• Evidence is now interaction-driven: click any line data point on the charts to open ranked articles/papers and related Fed supplemental factors for that point.