EUR/USD CCI research data — January–February 2024

Actual unchanged Forexite MetaStock quote records; no generated prices.
60 original daily archives were checked against their SHA-256 and ZIP CRC
before extraction. Source URLs, names and hashes are in research/sources.json.
The original archives are not duplicated here. The unchanged used minute
records are included with original CET and converted UTC. No Ask, spread,
execution, intraminute path or trading return is inferred.

Offline reproduction, Python 3 standard library only:
  python3 reproduce.py
  python3 verify_calculation.py
The first command recreates research/bars.json and eurusd-h4-cci.csv
byte-for-byte. The second independently groups the minute records and checks
CCI using direct 60-digit Decimal arithmetic, with tolerance 2e-11.

CCI: TP=(H+L+C)/3; average the latest N TP values. MD is the average
absolute distance of every TP in the SAME window from its CURRENT common
mean. CCI=(current TP-mean)/(.015*MD). N=14,20,50. If all N TP are
identical, MD=0 and this implementation explicitly returns CCI=0.
The producer uses exact integer-centered four-decimal HLC before division.
CCI is dimensionless and unbounded. There are no modeled trades here.

Timestamp grouping: record timestamps in (end-4h,end], labelled by end
in UTC. Winter CET minus one hour. No missing prices filled. Short
market-opening/closing groups remain; the source-end-cut group is omitted.
A broker's native H4 candle convention may differ.

Optional full source re-extraction: download the 60 named original ZIPs
to raw/, then run python3 prepare_data.py. The included manifest is used.
Alternatively set FBR_FOREXITE_ARCHIVES to their directory.
Optional SVG chart generation: python3 make_charts.py; the social PNG
export requires Inkscape. No network is needed for the offline replay.
