f6e418e400
Some checks failed
CI / lint (push) Failing after 28s
Deploy / notebooks (push) Has been skipped
Deploy / zotero (push) Has been skipped
Deploy / docs (push) Has been skipped
Deploy / api (push) Has been skipped
Deploy / mc (push) Has been skipped
Infra CI / notebooks (push) Failing after 13s
Infra CI / zotero (push) Successful in 12s
Infra CI / docs (push) Successful in 14s
Infra CI / api (push) Successful in 21s
Infra CI / mc (push) Successful in 12s
Deploy / report (push) Successful in 11s
CI / test (push) Has been cancelled
Investigation: characterized the consistent ~11K-12K ground-truth-only
rows per year (2016-2026) by joining against pfs.rvu. Result: 100% of
the gap is status_code='R'. Zero are A/T (no real calc bugs hidden in
the noise), zero are missing from pfs.rvu entirely.
Status R = Restricted Coverage. CMS applies special coverage rules
(certain settings, modifier requirements, NCD-driven) but the price
formula is identical to A: RVU × GPCI × CF. The pricer was filtering
status_code IN ('A', 'T'), treating R as if it were carrier-priced.
Fix: add 'R' to _ALGORITHMIC_STATUS. One-character change in the
SQL filter, plus updated docstrings.
Verified: live `stack rec pfs` for every year 2016-2026 now reports:
Ground-truth only: 0 (was 9,990 - 12,650)
Status: 100.00% exact
Residual deltas are single-digit ≤1¢ float-rounding noise across
~11M rows. Closes the framework's empirical concordance gap → #340
acceptance criteria fully satisfied.
Test: added 66666 (status R) to the in-memory fixture; updated count
assertions across the existing test class to reflect the new
4-row/3-match baseline; added test_includes_restricted_coverage
regression guard.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>