Forex strategy research: public aggregate and code package

Seven independent pair-level experiments, two existing rules, three evaluation
splits and 0/1/3 bp one-way hypothetical costs. This is not a combined portfolio.
Results derive from HistData Generic ASCII M1 inputs, cleaned/aggregated in the
local PostgreSQL forex_research project. Source https://www.histdata.com/f-a-q/.
2023 coverage is limited; inactivity is not evidence of successful risk control.
These idealized bid/reciprocal-price simulations omit financing, broker execution
and margin modelling. The extension followed review of the EURUSD findings.

Included: unchanged strategy producer code, original recorded-row indicator and
accounting definitions, derived summary/annual/coverage CSVs and browser daily
equity/drawdown indices. No original/normalized source prices, minute files,
frozen H1/ledger input corpus, credentials, native workbook or full notebook.

Aggregate/plot inspection is reproducible from the supplied outputs. Full strategy
replay requires your own authorized local HistData/PostgreSQL input snapshots;
the code expects eight named inputs under quality/multi_pair_access/results.
It is source-only for that full replay, not a self-contained backtest package.
On a separate authorized local research copy with those inputs, Python3.12+:
  python -m src.build_multi_pair_research --regression-only
  python -m src.build_multi_pair_research
The optional Tableau reshape expects real indicator outputs generated locally:
  python quality/reshape_multi_pair_explorer.py
No new open-source licence grant is asserted for the user-authored project code.
Raw-provider redistribution permission is not implied by this package.