How this platform works
Every index — flagship, showcase, or benchmark — is defined by a single YAML
file under backend/indexes/. One engine reads the definition, resolves a
universe, applies screening rules, computes weights, backtests, and exports
JSON + PDF artifacts that this site reads at build time. There is no
database, no server, and no auth on the public site — publishing an index
means editing YAML and pushing to git.
Index tiers
- Flagship — full methodology depth: an optimisation-based low-carbon index with factor regression, attribution, and sensitivity diagnostics.
- Showcase — simpler weighting schemes (equal, market-cap, dividend yield) proving the engine generalises beyond the flagship.
- Benchmark — a reference index other definitions can point to via
benchmark:/universe.reference_index:.
Data
Prices and metadata come from a free-tier provider fallback chain (yfinance → FMP → Alpha Vantage), cached locally. ESG/carbon data — where no free API exists — comes from manually curated, publication-dated CSVs so point-in-time backtests never leak information the market didn't have yet.
See docs/data_sources.md
and docs/limitations.md
for the full picture, including what's still pending.