Repository navigation
fix(schedule13): OwnershipComparison uses total_shares' max; add aggregation_basis - #1400
Merged
Merged
Conversation
…egation_basis (qsk4) OwnershipComparison summed aggregate_amount / percent_of_class over every reporting person, including rows flagged is_aggregate_exclude_shares, while total_shares / total_percent and _reported_total take the max over the unflagged persons: one rule, two implementations, disagreeing twice. In a control chain (fund, GP, manager, individual each reporting the same position) the sum counts one position once per person; the Jushi 13G fixture's two 10,000,000-share rows compared as 20,000,000. The comparison now goes through models._reported_total (reported_*) and an unreported-as-0 max (shares_change, percent_change, get_summary). The reported_* accessors are unreleased (#1379, after 5.59.1); shares_change's numbers change in 5.x as a correctness fix. The "max() is always correct" docstrings were false. Sampled joint filings: most rows restate one position, but GAMCO/Nevro (0000807249-25-000053) sub-advisers hold separate accounts, 1,322,950 by max vs 2,123,900 stated in Item 5(a), and the 33-person ZIM group (0001178913-25-004165) matches neither sum nor max. No structured field says which rows roll up into which, so a new aggregation_basis reports the shape: 'single', 'identical', 'parent_sums_children' (largest row = sum of the distinct smaller ones; Blackstone 0000950170-25-090888 4,782,781 = 2,989,238 + 1,793,543) or 'ambiguous'. On 60 joint filings from 2025Q3: 13D 15 identical / 13 ambiguous / 2 single; 13G 23 identical / 3 parent / 4 ambiguous. to_context() flags ambiguous totals. The guide's "Joint vs. Separate Filers" section described member_of_group / >100% / identical- count detection that was never implemented; rewritten. The Item 5(a) stated total is bead edgartools-vazz. 1379's tests pinned the summed values; updated to the max (Aadi 4,535,000 -> 2,435,000, Jushi 20,000,000 -> 10,000,000). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
dgunning
enabled auto-merge (squash)
October 1, 2026 23:00
Merged
dgunning
added a commit
that referenced
this pull request
Oct 20, 2026
…edgartools-u6tkw) (#1621) * docs(upgrade): pre-rc review of the 6.0 migration guide Every breaking PR since v5.62.0 (17) and every pending removed/changed fragment (19) has a section; nothing was missing. Fixes from reading the whole guide and running all 43 python blocks against main: - The intro said 6.0 is still in development; it now says the page covers every change merged for 6.0. - "Each of these warns in 5.x" was false for two removals that never warned (edgar.core.Result, the names outside edgar.__all__); the note now lists what warned and names the two that did not. - "Still to come": public API enforcement shipped (#1611) and the source tree reorg was deferred to 6.x; both rows are gone. The Schedule 13D/G row stays until edgartools-u6tkw lands. - The five `sed -i ''` rewrites were macOS-only; they are perl -pi -e now, tested on a sample file. - The get_concept() lease example returned None for every company: get_concept takes synonym-group names, not raw tags. It uses get_fact() with a same-period check (Snowflake: 59.8M + 420.0M = 479.8M). - Smaller accuracy fixes: the access modes did not move to edgar.settings, __all__ shipped in 5.47.0, the Apple blank-cell row is fiscal 2023, the Citigroup Item 1 count names its filing, the mda example's comment. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * feat(schedule13)!: figures a filing does not report are None, not 0 (edgartools-u6tkw) 6.0 flip staged in 5.x by #1379/#1384/#1400. ReportingPerson's six ownership figures hold None for an unreported figure (__post_init__ keeps the fields and unreported_fields in step, so 5.x-style construction with a placeholder 0 still works); total_voting_power/total_dispositive_power are None when either half is; Schedule13D/G total_shares/total_percent use the shared _reported_total rule (None for a counted person leaving the figure out or no reporting-person rows; 0 when every person is excluded). OwnershipComparison.shares_change/percent_change and get_summary() values are None for unreported figures and no longer emit a FutureWarning; reported_shares_change/reported_percent_change and ReportingPerson.reported() stay as non-warning aliases. Tests in test_issue_1379 that asserted the 5.x placeholder 0s and the FutureWarning now assert None and no warning. New regression test pins values from BML Investment Partners' 13D on Aadi Bioscience, 0001373604-24-000118. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * docs(upgrade): Schedule 13D/G unreported figures section; Still to come becomes Not in 6.0 (edgartools-u6tkw) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Bead edgartools-qsk4. Decision recorded on the bead: use max plus an
aggregation_basisflag, shipping in 5.x.Problem
OwnershipComparisonsummed every reporting person's shares and percent, including rows flaggedis_aggregate_exclude_shares. Meanwhiletotal_sharesandtotal_percenttake the max over the unflagged persons. In a control chain, where a fund, its GP and its manager each report the same position, the sum counts that position once per person. The Jushi 13G fixture's two 10,000,000-share rows compared as 20,000,000.The "max() is always correct" docstrings were also false. In a sample of joint filings:
0000807249-25-000053): sub-advisers hold separate accounts, so the max is 1,322,950 while Item 5(a) states 2,123,900 shares (5.54%).0001178913-25-004165, 33 persons): matches neither the sum nor the max.No structured field says which rows roll up into which.
Changes
Comparison:
OwnershipComparisonnow uses the same max astotal_shares.reported_*goes throughmodels._reported_total;shares_change,percent_changeandget_summary()use an unreported-as-0 max. Thereported_*accessors are unreleased (fix(schedule13): no sale or purchase from ownership figures a filing leaves out #1379 came after 5.59.1).shares_change's numbers change in 5.x as a correctness fix; the changelog says so.aggregation_basison Schedule13D and Schedule13G:singleidenticalparent_sums_children0000950170-25-090888: 4,782,781 = 2,989,238 + 1,793,543)ambiguoustotal_sharesto_context()flags ambiguous totals.Docs: the false docstrings are fixed. The guide's "Joint vs. Separate Filers" section described detection logic that was never implemented, and it's rewritten. The ownership skill gets a Schedule 13 pattern.
Distribution on 60 joint filings from 2025 Q3:
Reading the filer's own Item 5(a) total is bead edgartools-vazz.
Verification
ambiguous, flagged in context); the Jushi fixture (identical).ambiguous, 1,322,950) and Blackstone (parent_sums_children, 4,782,781).test-fast: the only failure is the dt1f1 20-F test, which also fails on main.🤖 Generated with Claude Code