GBP/USD KAMA calculation research data

Data: Forexite daily MetaStock quote archives, Jan-Feb 2024.
The included M1 rows preserve the published OHLC and source CET timestamps.
Original daily ZIP files are not bundled; URLs and original SHA-256 hashes
are provided in research/sources.json. Those original bytes were checked
before extraction. No missing prices, Ask quotes, spreads or fills are inferred.

Offline replay (Python 3 standard library only):
  python3 reproduce.py
  python3 verify_calculation.py

The first command recreates bars.json and gbpusd-h4-kama.csv from the
included unchanged minute rows. The second independently checks all H4
groups and uses Decimal precision 50 to check KAMA/EMA calculations.
The verification tolerance is 2e-12; displayed guide values are rounded.

KAMA: N = 10, 20 or 30; fast=2, slow=30. SMA of first N closes is
the seed at index N-1; the first ER/update is at index N. Zero path => ER=0.
KAMA uses the complete preceding dataset, not a visible chart segment.
EMA(10) is seeded with the first 10-close SMA at index 9.

H4 grouping: published M1 record timestamps in (end-4h,end], labelled
by end at 00/04/08/12/16/20 UTC. CET minus 1 hour in this winter interval.
This declared grouping may differ from broker-native candles. Short
opening/closing groups are retained, final cut group omitted. No filling.

Optional chart generation:
  python3 make_charts.py
SVG writing uses the standard library. PNG export needs Inkscape installed.
The educational charts are historical observations, not a trading backtest.
