Stage E (writing): main.tex v1.1 / November 2026, abstract rewritten to the audited wording (FIRC0001-THEO-18, ADVE-04, ADVE-05, EDIT-07, ADVE-09), PDF metadata, cleveref; provisional \providecommand macros for Stage D audit numbers (sections/numbers_pending.tex)

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Simon-Pierre Boucher committed yesterday (Nov 5, 2026) parent 9aef075 Browse files

2 changed files +408 −3

modified paper/main.tex +18 −3

				`@@ -52,6 +52,7 @@`
52	urlcolor = NavyBlue,	52	urlcolor = NavyBlue,
53	breaklinks = true	53	breaklinks = true
54	]{hyperref}	54	]{hyperref}
		55	+\usepackage[capitalise,noabbrev]{cleveref} % \cref/\Cref in the Stage E prose (loaded after hyperref)
55		56
56	% --- Bibliography ---	57	% --- Bibliography ---
57	\usepackage[authoryear,round,semicolon]{natbib}	58	\usepackage[authoryear,round,semicolon]{natbib}
				`@@ -93,6 +94,7 @@`
93	\usepackage{adjustbox}	94	\usepackage{adjustbox}
94	\usepackage{upquote} % straight quotes in verbatim (extraction contract, gate regex)	95	\usepackage{upquote} % straight quotes in verbatim (extraction contract, gate regex)
95	\input{tables/numbers} % generated inline-number macros (code/make_tables.py); never edit by hand	96	\input{tables/numbers} % generated inline-number macros (code/make_tables.py); never edit by hand
		97	+\input{sections/numbers_pending} % PROVISIONAL \providecommand macros for Stage D audit numbers not yet emitted by the generator (Agent 10); each line names its results/ source
96		98
97	% ============================================================================	99	% ============================================================================
98	% METADATA — Edit these fields for each working paper	100	% METADATA — Edit these fields for each working paper
				`@@ -100,11 +102,16 @@`
100	\newcommand{\WPnumber}{0001}	102	\newcommand{\WPnumber}{0001}
101	\newcommand{\WPtitle}{From Corporate Disclosures to a Russell~3000 Strategic Project Database}	103	\newcommand{\WPtitle}{From Corporate Disclosures to a Russell~3000 Strategic Project Database}
102	\newcommand{\WPsubtitle}{Identification, Measurement, Validation, and Research Applications}	104	\newcommand{\WPsubtitle}{Identification, Measurement, Validation, and Research Applications}
103	−\newcommand{\WPdate}{October 28, 2026}	105	+% Stage E (2026-11-05), FIRC0001-THEO-18 / cluster I: the Chief Editor noted that "Identification" in the
104	−\newcommand{\WPversion}{1.0 (working draft under validation)}	106	+% subtitle reads as econometric identification to a finance audience and that the estimand is a Russell 3000
		107	+% proxy (iShares IWV holdings). Proposed alternative, NOT imposed (author decision):
		108	+% \newcommand{\WPsubtitle}{Construction, Measurement, Validation Status, and Research Applications}
		109	+% The title is unchanged; the abstract and Section 3 state the proxy explicitly.
		110	+\newcommand{\WPdate}{November 2026}
		111	+\newcommand{\WPversion}{1.1 (revised after peer review; research database under validation)}
105	\newcommand{\WPfunding}{Research funded by the FIRC (Fonds institutionnel de recherche et de création), Université du Québec en Outaouais.}	112	\newcommand{\WPfunding}{Research funded by the FIRC (Fonds institutionnel de recherche et de création), Université du Québec en Outaouais.}
106	\newcommand{\WPabstract}{%	113	\newcommand{\WPabstract}{%
107	− We document the construction of SPID, a project-level research database that reconstructs the strategic initiatives disclosed by U.S.-listed companies in their SEC filings. Starting from an issuer universe that merges a current S\&P~500 snapshot with the holdings of the iShares Russell~3000 ETF (2{,}564 distinct SEC registrants), the pipeline collects every 10-K, 10-Q and 8-K filed since January 2010 (527{,}548 documents), segments them into 1.34 million sections, and submits 1.12 million eligible sections to a schema-constrained large-language-model extractor. The result is 308{,}065 evidence-linked project mentions, resolved into 121{,}099 candidate projects for 2{,}492 issuers, each tied to its SEC accession number, section and source passage, together with a controlled 21-category taxonomy, status, cited years, locations, technologies, partners and disclosed amounts. We describe the universe definition and its reconciliation, the distributed acquisition and parsing architecture, the extraction contract and its confidence semantics, the mention-to-project resolution rule and its known failure modes, and the point-in-time structure of the timelines. We then set out a validation framework---stratified human annotation, precision/recall on an independent gold standard, linkage audits and coverage gates---and state explicitly which of its components are implemented and which remain to be executed. Descriptive patterns are reported for the full corpus, with the caveat that a disclosed project is neither a verified investment nor a realized expenditure. The database should be read as a research instrument under validation rather than as an exhaustive census of Russell~3000 projects.%	114	+ Can strategic projects be identified, classified, dated and linked across the SEC filings of a broad cross-section of U.S.-listed issuers? We document SPID, a project-level research database built for a Russell~3000 proxy (iShares IWV holdings, \SPIDrussellDistinctCiks\ of \SPIDrussellIndexConstituents\ index constituents) merged with the S\&P~500: \SPIDfilingsTotal\ Forms 10-K, 10-Q and 8-K of \SPIDuniverseN\ registrants filed since 2010, segmented into \SPIDsectionsTotal\ sections, of which \SPIDgatePassTotal\ reached a large-language-model extractor in JSON mode. The output is \SPIDmentions\ section-linked candidate mentions and \SPIDprojectCountCurrentKey\ candidate projects under the current resolution key (\SPIDprojectsKeyTypeDropped--\SPIDprojectsKeyStrict\ under alternative keys). Mechanical audits find \SPIDgroundAllPct\% of structured fields verbatim in the source section, a median \SPIDmdnaShownMedianPct\% of a 10-K MD\&A shown to the model, and \SPIDfinMentionsPct\% of mentions describing financing instruments. Precision and recall against human labels are not yet measured; the annotation frame is delivered. SPID is a research instrument under validation, not a census of projects.%
108	}	115	}
109	\newcommand{\WPkeywords}{SEC filings, text as data, corporate investment, strategic projects, information extraction, large language models, entity resolution, Russell 3000, measurement validation}	116	\newcommand{\WPkeywords}{SEC filings, text as data, corporate investment, strategic projects, information extraction, large language models, entity resolution, Russell 3000, measurement validation}
110	\newcommand{\WPjel}{C55, C81, G31, G38, M41, O33}	117	\newcommand{\WPjel}{C55, C81, G31, G38, M41, O33}
				`@@ -126,6 +133,14 @@`
126	Gatineau, Québec, Canada J9A 1L8%	133	Gatineau, Québec, Canada J9A 1L8%
127	}	134	}
128		135
		136	+% --- PDF metadata (Part X.3) ---
		137	+\hypersetup{
		138	+ pdftitle = {\WPtitle: \WPsubtitle},
		139	+ pdfauthor = {Manel Kammoun; Djerrycharli Tandjambianda; Simon-Pierre Boucher},
		140	+ pdfsubject = {UQO Working Paper No. \WPnumber, version \WPversion. JEL: \WPjel},
		141	+ pdfkeywords = {\WPkeywords}
		142	+}
		143	+
129	% ============================================================================	144	% ============================================================================
130	% DOCUMENT	145	% DOCUMENT
131	% ============================================================================	146	% ============================================================================

added paper/sections/numbers_pending.tex +390 −0

++% ----------------------------------------------------------------------------
++% PROVISIONAL inline-number macros for the Stage E prose (Agent 10, 2026-11-05).
++% Every value below is copied from a Stage D audit output under results/ (file and
++% field given on each line) and is defined with \providecommand so that the
++% generated definition in tables/numbers.tex (code/make_tables.py, Agent 9) wins
++% as soon as it exists. The same list, with value/source/formula, is filed for the
++% generator in improvements/FIRC-0001/macro_requests_writing.json. Delete this
++% file once make_tables.py emits every macro (check: grep providecommand).
++% ----------------------------------------------------------------------------
++
++% ---- ledger, calls, outage (results/ledger/*.json, gate_pass_by_run.csv) ----
++\providecommand{\SPIDoutageRowsRemoved}{149{,}352}            % log_summary.json:october.outage_rows_reconciliation.difference
++\providecommand{\SPIDoutageLargestGapStart}{01:25}            % throughput.json:largest_gaps_seconds[-1] (2026-10-28)
++\providecommand{\SPIDoutageLargestGapEnd}{06:40}              % throughput.json:largest_gaps_seconds[-1]
++\providecommand{\SPIDoutageZeroBatchRange}{47--60}            % log_summary.json:october.anomalous_batches (batch ids with 0 mentions)
++\providecommand{\SPIDoutageZeroBatchWindow}{01:44--06:05}     % log_summary.json:october.anomalous_batches (first/last timestamp)
++\providecommand{\SPIDoutageSurvivingRows}{504}                % backfill_block.json:rows_in_outage_window[0]
++\providecommand{\SPIDoutageSurvivingMentions}{1{,}011}        % backfill_block.json:rows_in_outage_window[1]
++\providecommand{\SPIDresumeBatchSections}{5{,}198}            % log_summary.json:october.resume_batch1.log_sections
++\providecommand{\SPIDresumeBatchCiks}{20}                     % log_summary.json:october.resume_batch1.ciks (length)
++\providecommand{\SPIDresumeBatchCalls}{2{,}686}               % log_summary.json:october.resume_batch1.ledger_calls
++\providecommand{\SPIDresumeBatchProductive}{23}               % log_summary.json:october.resume_batch1.ledger_with_mentions
++\providecommand{\SPIDresumeBatchExpectedProductive}{536}      % resume_batch1_vs_expected.csv: sum(expected_with_mentions) = 536.2
++\providecommand{\SPIDresumeBatchMentions}{34}                 % log_summary.json:october.resume_batch1.log_mentions
++\providecommand{\SPIDresumeBatchWindow}{06:38--06:40}         % log_summary.json:october.resume_batch1.window (2026-10-28)
++\providecommand{\SPIDresumeBatchZeroSharePct}{99.2}           % flush_zero_groups_after_0622.csv (06:40 minute bin, 99.15)
++\providecommand{\SPIDjuneCalls}{95{,}286}                     % gate_pass_by_run.csv:backfilled_june.calls
++\providecommand{\SPIDjuneProductiveShareCalls}{25.6}          % gate_pass_by_run.csv:backfilled_june.productive_pct_calls
++\providecommand{\SPIDoctoberProductiveShareCalls}{23.6}       % gate_pass_by_run.csv:october.productive_pct_calls
++\providecommand{\SPIDlogExtractionCompleteSum}{1{,}117{,}479} % log_summary.json:october.extraction_complete_sum_sections
++\providecommand{\SPIDlogPendingSecondStart}{968{,}127}        % log_summary.json:october.pending_lines[1]
++\providecommand{\SPIDconfExactDefault}{3{,}278}               % code_vs_data_checks.json:confidence_exact_0600_total
++\providecommand{\SPIDconfExactDefaultJune}{304}               % code_vs_data_checks.json:confidence_exact_0600_by_run
++\providecommand{\SPIDconfExactDefaultOctober}{2{,}974}        % code_vs_data_checks.json:confidence_exact_0600_by_run
++\providecommand{\SPIDamountRawDeviations}{584}                % code_vs_data_checks.json:amount_raw_rule_mismatch
++\providecommand{\SPIDamountNegative}{24}                      % results/grounding/amount_tail.json:lt_0
++\providecommand{\SPIDamountZero}{584}                         % results/grounding/amount_tail.json:eq_0
++\providecommand{\SPIDledgerBackendValue}{llm}                 % code_vs_data_checks.json:distinct_backend_ledger
++
++% ---- grounding (results/grounding/grounding_by_group.csv, grounding_all_structured.csv, grounding_element_level.csv, fallback_bounds.json, amount_tail.json) ----
++\providecommand{\SPIDgroundBootstrapReps}{1{,}000}            % a02_grounding.py (bootstrap replicates, clustered by section)
++\providecommand{\SPIDgroundYearsPct}{99.8}                    % grounding_by_group.csv all/years 0.9976
++\providecommand{\SPIDgroundYearsCiLow}{99.7}                  % ci_low 0.9974
++\providecommand{\SPIDgroundYearsCiHigh}{99.8}                 % ci_high 0.9978
++\providecommand{\SPIDgroundPartnersPct}{99.4}                 % all/partners 0.9940
++\providecommand{\SPIDgroundAmountNumberPct}{93.7}             % all/amount_number 0.9372
++\providecommand{\SPIDgroundAmountCurrencyPct}{80.5}           % all/amount_with_currency 0.8047
++\providecommand{\SPIDgroundLocationsPct}{78.3}                % all/locations 0.7832
++\providecommand{\SPIDgroundLocationsTenK}{85.7}               % form/locations/10-K 0.8571
++\providecommand{\SPIDgroundLocationsTenQ}{82.1}               % form/locations/10-Q 0.8207
++\providecommand{\SPIDgroundLocationsEightK}{67.8}             % form/locations/8-K 0.6780
++\providecommand{\SPIDgroundTechPct}{63.6}                     % all/technologies 0.6358
++\providecommand{\SPIDgroundElemYears}{99.7}                   % grounding_element_level.csv years
++\providecommand{\SPIDgroundElemPartners}{99.6}                % partners
++\providecommand{\SPIDgroundElemLocations}{84.2}               % locations
++\providecommand{\SPIDgroundElemTech}{70.6}                    % technologies
++\providecommand{\SPIDgroundAllPct}{88.2}                      % grounding_all_structured.csv all 88.17
++\providecommand{\SPIDgroundAllN}{276{,}168}                   % grounding_all_structured.csv all n_with_fields
++\providecommand{\SPIDgroundAllTenK}{92.2}                     % 10-K 92.15
++\providecommand{\SPIDgroundAllTenQ}{92.4}                     % 10-Q 92.41
++\providecommand{\SPIDgroundAllEightK}{76.6}                   % 8-K 76.57
++\providecommand{\SPIDgroundAllJune}{84.5}                     % backfilled_june 84.51
++\providecommand{\SPIDgroundAllOctober}{88.7}                  % october 88.74
++\providecommand{\SPIDgroundAllTopBin}{86.4}                   % 0.9-1.0 bin 86.40
++\providecommand{\SPIDgroundAllLowerBinsRange}{90.7--94.2}     % bins 0.45-0.6 ... 0.8-0.9
++\providecommand{\SPIDyearsNonEmptyN}{252{,}469}               % fallback_bounds.json:years_nonempty
++\providecommand{\SPIDyearsRegexEqualN}{98{,}438}              % fallback_bounds.json:years_eq_regex_desc
++\providecommand{\SPIDyearsRegexEqualPct}{39}                  % 98,438 / 252,469
++\providecommand{\SPIDtechNonEmptyN}{36{,}740}                 % fallback_bounds.json:techs_nonempty
++\providecommand{\SPIDtechRegexEqualN}{16{,}949}               % fallback_bounds.json:techs_eq_regex_desc
++\providecommand{\SPIDtechRegexEqualPct}{46}                   % 16,949 / 36,740
++\providecommand{\SPIDyearsAbsentFromText}{604}                % fallback_bounds.json:years_not_subset_of_text
++\providecommand{\SPIDtypeSignalMaPct}{79.8}                   % fallback_bounds.json:type_signal_in_name_desc_by_type ma_integration
++\providecommand{\SPIDtypeSignalTechDeployPct}{5.4}            % technology_deployment 5.35
++\providecommand{\SPIDtechDeployMentionedN}{124}               % fallback_bounds.json:tech_deploy_total[1]
++\providecommand{\SPIDtechDeployMentionedNoSignal}{90}         % fallback_bounds.json:tech_deploy_mentioned_no_any_signal
++\providecommand{\SPIDamountMaxSci}{1.0 \times 10^{12}}        % amount_tail.json:max_amount (math mode)
++\providecommand{\SPIDamountGeFiftyBMentions}{161}             % amount_tail.json:ge_50e9
++\providecommand{\SPIDamountGeFiftyBProjects}{104}             % amount_tail.json:projects_max_amount_ge_50e9
++\providecommand{\SPIDamountTopNumberFound}{191}               % amount_tail.json:top200_grounded_number[0]
++\providecommand{\SPIDamountTopCurrencyFound}{178}             % amount_tail.json:top200_grounded_number[1]
++\providecommand{\SPIDamountTopN}{200}                         % amount_tail.json:top200_grounded_number[2]
++\providecommand{\SPIDamountNonUsdTokenN}{1{,}837}             % amount_tail.json:nonusd_within_40_share[0]
++\providecommand{\SPIDamountLocatedN}{112{,}692}               % amount_tail.json:nonusd_within_40_share[1]
++\providecommand{\SPIDamountNonUsdTokenPct}{1.6}               % 1,837 / 112,692
++
++% ---- resolver (results/resolver/composition.json, spec_curve.csv, top_name_tokens.csv, overmerge_by_type.csv, ties_idempotence.json, june_october_reconciliation.json) ----
++\providecommand{\SPIDmentionsKeyedLocation}{118{,}783}        % composition.json:mentions_by_key_branch location
++\providecommand{\SPIDmentionsKeyedNameToken}{187{,}212}       % composition.json:mentions_by_key_branch name_token
++\providecommand{\SPIDmentionsKeyedNameTokenPct}{60.8}         % 187,212 / 308,065
++\providecommand{\SPIDtopTokenMentions}{40{,}407}              % top_name_tokens.csv acquisition
++\providecommand{\SPIDtopTokenProjects}{2{,}256}               % top_name_tokens.csv acquisition
++\providecommand{\SPIDtopTokenSharePct}{21.4}                  % pct_of_tokenkeyed_mentions
++\providecommand{\SPIDlocationKeysCountry}{8{,}126}            % composition.json:location_keyed_projects_by_geo_kind country
++\providecommand{\SPIDlocationKeysUsState}{6{,}110}            % us_state
++\providecommand{\SPIDlocationKeysCountryStatePct}{26}         % (8,126 + 6,110) / 54,658
++\providecommand{\SPIDlocationKeyUnitedStatesProjects}{2{,}551}% top_location_keys.csv 'united states'
++\providecommand{\SPIDprojectsKeyCikType}{21{,}027}            % spec_curve.csv k_cik_type 0.45
++\providecommand{\SPIDprojectsKeyTypeDropped}{96{,}749}        % k_type_dropped 0.45
++\providecommand{\SPIDprojectsKeyNameOnly}{114{,}244}          % k_name_only 0.45
++\providecommand{\SPIDprojectsKeyCityNorm}{123{,}246}          % k_city_normalised 0.45
++\providecommand{\SPIDprojectsKeyStopwords}{130{,}101}         % k_current_stopwords 0.45
++\providecommand{\SPIDprojectsKeyTwoTokens}{145{,}697}         % k_two_tokens 0.45
++\providecommand{\SPIDprojectsKeyStrict}{198{,}488}            % k_strict 0.45
++\providecommand{\SPIDprojectsKeyMentionLevel}{308{,}065}      % k_mention_level 0.45
++\providecommand{\SPIDprojectsCurrentKeyAtNine}{83{,}483}      % k_current 0.9
++\providecommand{\SPIDsingleMentionShareStrict}{75.8}          % k_strict 0.45 single_mention_pct
++\providecommand{\SPIDsingleMentionShareCikType}{20.7}         % k_cik_type 0.45
++\providecommand{\SPIDchenierClusterCurrent}{8}                % example_cluster_size_cheniere k_current
++\providecommand{\SPIDchenierClusterStrict}{1}                 % k_strict
++\providecommand{\SPIDchenierClusterNameOnly}{40}              % k_name_only
++\providecommand{\SPIDchenierClusterCikType}{176}              % k_cik_type
++\providecommand{\SPIDmaGeThreeNamesPct}{14.6}                 % overmerge_by_type.csv ma_integration pct_ge3_names
++\providecommand{\SPIDmaGeThreeNamesOverTenPct}{86.6}          % overmerge_ma_by_bucket.csv >10 mentions
++\providecommand{\SPIDmaGeThreePartnersPct}{10.9}              % overmerge_by_type.csv ma_integration pct_ge3_partners_among_with_partner
++\providecommand{\SPIDmaAmountOtherCounterpartyN}{1{,}630}     % composition.json:amount_from_other_counterparty_ma[0]
++\providecommand{\SPIDmaAmountOtherCounterpartyDenom}{12{,}372}% composition.json:amount_from_other_counterparty_ma[1]
++\providecommand{\SPIDamountOtherCounterpartyN}{2{,}347}       % composition.json:amount_from_other_counterparty[0]
++\providecommand{\SPIDamountOtherCounterpartyDenom}{21{,}908}  % composition.json:amount_from_other_counterparty[1]
++\providecommand{\SPIDprojectsTieDistinctDescriptions}{29{,}954} % ties_idempotence.json
++\providecommand{\SPIDrebuiltNameDiff}{394}                    % ties_idempotence.json:rebuilt_vs_stored_diff[0]
++\providecommand{\SPIDrebuiltDescDiff}{600}                    % ties_idempotence.json:rebuilt_vs_stored_diff[2]
++\providecommand{\SPIDtiebreakNameDiff}{11{,}484}              % ties_idempotence.json:deterministic_vs_stored_diff[0]
++\providecommand{\SPIDtiebreakStatusDiff}{503}                 % [1]
++\providecommand{\SPIDtiebreakDescDiff}{18{,}256}              % [2]
++\providecommand{\SPIDjuneProjects}{19{,}227}                  % june_october_reconciliation.json:projects_from_backfilled_block_only
++\providecommand{\SPIDspMentions}{70{,}367}                    % sp500_mentions_total
++\providecommand{\SPIDspProjects}{27{,}766}                    % sp500_projects_total
++\providecommand{\SPIDspProjectsJuneOnly}{15{,}910}            % sp500_projects_by_run_composition
++\providecommand{\SPIDspProjectsBothRuns}{3{,}317}
++\providecommand{\SPIDspProjectsOctoberOnly}{8{,}539}
++\providecommand{\SPIDoctoberSpTenQSections}{116{,}231}        % sp500_october_ledger_by_form 10-Q
++\providecommand{\SPIDoctoberSpTenQFilings}{19{,}399}
++\providecommand{\SPIDoctoberSpTenQMentions}{28{,}963}         % sp500_mentions_by_run_form october 10-Q
++\providecommand{\SPIDjuneTenQFilings}{3{,}710}                % june_10q_accessions_in_block_ledger
++\providecommand{\SPIDjuneTenQSections}{23{,}873}              % ledger_block_accessions_by_form 10-Q
++\providecommand{\SPIDjuneTenQMentions}{5{,}817}               % sp500_mentions_by_run_form backfilled_june 10-Q
++\providecommand{\SPIDoctoberSpTenKMentions}{910}              % october 10-K
++\providecommand{\SPIDoctoberSpEightKMentions}{564}            % october 8-K
++
++% ---- two runs / two segmenters (results/runs/git_evidence.md, segmenters.json, segmenter_item_comparison.csv, funnel_10k.json, within_*.csv, table6_cohort*.json/csv, fiscal_year_audit.json) ----
++\providecommand{\SPIDrepoCommits}{six}                        % git_evidence.md: all commits dated 2026-10-27/28
++\providecommand{\SPIDextractorMtimeDate}{June~17, 2026}       % git_evidence.md: llm.py, rules.py, base.py, taxonomy.py, resolve.py, ingest.py mtimes
++\providecommand{\SPIDdiffStoreLines}{$+3/-1$}                 % git_diff_df7905d_2324b58.patch (store.py)
++\providecommand{\SPIDdiffParserLines}{$+13/-7$}               % git_diff_0304133_2324b58.patch (filing_parser.py)
++\providecommand{\SPIDlegacyTenKRows}{27{,}164}                % segmenters.json:legacy_rows
++\providecommand{\SPIDlegacyTenKFilings}{7{,}368}              % segmenters.json:legacy_accessions
++\providecommand{\SPIDlegacyShareTenKLedgerPct}{18.9}          % 27,164 / 143,704
++\providecommand{\SPIDsegPairs}{25{,}673}                      % segmenters.json:legacy_rows_matched_by_item
++\providecommand{\SPIDlegacyMdnaDollar}{141}                   % segmenter_item_comparison.csv md_and_a legacy_mean_dollar
++\providecommand{\SPIDparsedMdnaDollar}{272}                   % parsed_mean_dollar
++\providecommand{\SPIDlegacyMdnaWords}{9{,}921}                % legacy_median_words
++\providecommand{\SPIDparsedMdnaWords}{11{,}661}               % parsed_median_words
++\providecommand{\SPIDlegacyFewerMoneyPct}{92.3}               % pct_legacy_fewer_money_tokens
++\providecommand{\SPIDlegacyMdnaNumeric}{596}                  % legacy_mean_numeric
++\providecommand{\SPIDparsedMdnaNumeric}{1{,}177}              % parsed_mean_numeric
++\providecommand{\SPIDpropertiesLtHalfPct}{19.4}               % properties pct_legacy_lt_half_words
++\providecommand{\SPIDfullBodyMedianWords}{44{,}514}           % segmenters.json:full_body_rows[2]
++\providecommand{\SPIDfullBodyYield}{0.67}                     % full_body_ledger_yield[3]
++\providecommand{\SPIDfullBodyProductivePct}{25.7}             % full_body_ledger_yield[4]
++\providecommand{\SPIDfullBodyWithParsedItems}{1{,}117}        % implemented_changes.md N4 (segmenters.json)
++\providecommand{\SPIDfullBodyUnsubmittedSections}{5{,}385}    % implemented_changes.md N4 (segmenters.json)
++\providecommand{\SPIDfullDocumentTenK}{86}                    % segmenters.json / funnel_10k.json
++\providecommand{\SPIDfullDocumentTenKLegacy}{51}              % parsed_section_names_same_accessions full_document
++\providecommand{\SPIDfullDocumentTenKInvisible}{35}           % 86 - 51
++\providecommand{\SPIDfullDocumentTenQ}{243}                   % funnel_10k.json:parsed_10q_items full_document
++\providecommand{\SPIDfullDocumentTenQLedgered}{241}           % funnel_10k.json:ledger_10q_items full_document
++\providecommand{\SPIDtenKparsedRows}{341{,}207}               % funnel_10k.json:parsed_10k_rows_total
++\providecommand{\SPIDtenKoutOfScopeRows}{188{,}518}           % parsed_10k_rows_out_of_scope_item
++\providecommand{\SPIDtenKlegacyAccessionRows}{35{,}037}       % parsed_10k_in_scope_rows_of_legacy_accessions
++\providecommand{\SPIDtenKunderThirtyRows}{1{,}112}            % parsed_10k_in_scope_rows_lt30_words_non_legacy
++\providecommand{\SPIDtenKparsedLedgered}{116{,}540}           % parsed_10k_in_scope_rows_ge30_non_legacy
++\providecommand{\SPIDwithinProductiveJune}{12.4}              % within_sp500_by_run_form.csv backfilled_june 10-Q productive_pct
++\providecommand{\SPIDwithinProductiveOctober}{12.7}           % october 10-Q
++\providecommand{\SPIDwithinMentionsPerRowJune}{0.244}         % mentions_per_row
++\providecommand{\SPIDwithinMentionsPerRowOctober}{0.250}
++\providecommand{\SPIDwithinConfJune}{0.889}                   % within_sp500_10q_mentions_by_run.csv mean_conf
++\providecommand{\SPIDwithinConfOctober}{0.893}
++\providecommand{\SPIDwithinMaShareJune}{49.9}                 % within_sp500_10q_type_mix_by_run.csv ma_integration
++\providecommand{\SPIDwithinMaShareOctober}{45.7}
++\providecommand{\SPIDpairedIssuers}{81}                       % within_issuer_paired_10q.json:issuers_paired
++\providecommand{\SPIDpairedDeltaPp}{0.6}                      % mean_diff_productive_share_oct_minus_june (pp)
++\providecommand{\SPIDpairedSePp}{1.3}                         % se (pp)
++\providecommand{\SPIDcohortIssuersBoth}{491}                  % table6_cohort.csv both.issuers
++\providecommand{\SPIDcohortIssuersRussellOnly}{1{,}992}       % russell_only.issuers
++\providecommand{\SPIDcohortIssuersSpOnly}{9}                  % sp500_only.issuers
++\providecommand{\SPIDcohortBothPerIssuerAll}{55.8}            % both.per_issuer_all
++\providecommand{\SPIDcohortRussellOnlyPerIssuer}{46.9}        % russell_only.per_issuer_all
++\providecommand{\SPIDcohortBothPerIssuerOctober}{25.1}        % both.per_issuer_october_only
++\providecommand{\SPIDcohortOlsAll}{8.84}                      % table6_cohort_regression.json:projects_all_on_sp500.in_sp500.coef
++\providecommand{\SPIDcohortOlsAllSe}{1.62}                    % hc1_se
++\providecommand{\SPIDcohortOlsOctober}{-22.98}                % projects_october_only_on_sp500.in_sp500.coef
++\providecommand{\SPIDcohortOlsOctoberSe}{1.20}
++\providecommand{\SPIDcohortOlsControlled}{-0.28}              % projects_all_on_sp500_log_rows.in_sp500.coef
++\providecommand{\SPIDcohortOlsControlledSe}{1.52}
++\providecommand{\SPIDcohortOlsN}{2{,}499}                     % n
++\providecommand{\SPIDfiscalYearNullMentions}{68{,}827}        % fiscal_year_audit.json
++\providecommand{\SPIDfiscalYearNullOctTenK}{823}
++\providecommand{\SPIDfiscalYearNullOctTenQ}{28{,}074}
++\providecommand{\SPIDfiscalYearNullFilings}{134{,}124}        % src_filings_fiscal_year_null_by_form (sum)
++\providecommand{\SPIDfiscalYearFillAccessions}{31{,}976}      % fiscal_year_fill_by_accession.csv rows
++
++% ---- focus windows (results/focus/stats_keys.json, focus_coverage_by_form_item.csv, shown_fraction_bins.csv, mdna_10k_short_by_year.csv; results/rules/focus_recall_loss*.csv) ----
++\providecommand{\SPIDfocusOverBudgetSections}{326{,}095}      % stats_keys.json:ledger_sections_over_9000_chars
++\providecommand{\SPIDfocusOverBudgetSectionsPct}{29}          % 326,095 / 1,123,098
++\providecommand{\SPIDfocusOverBudgetCalls}{324{,}386}         % stats_keys.json:calls_over_9000_chars
++\providecommand{\SPIDfocusOverBudgetCallsPct}{46}             % 324,386 / 703,725
++\providecommand{\SPIDfocusMeanShownOverBudgetPct}{15.9}       % stats_keys.json:mean_shown_frac_over_budget
++\providecommand{\SPIDmdnaN}{28{,}919}                         % stats_keys.json:ledger_10k_mdna_n
++\providecommand{\SPIDmdnaMedianWords}{9{,}905}                % stats_keys.json:ledger_10k_mdna_median_words
++\providecommand{\SPIDmdnaMedianChars}{64{,}194}               % ledger_10k_mdna_median_chars
++\providecommand{\SPIDmdnaShownMedianPct}{9.3}                 % ledger_10k_mdna_shown_p50
++\providecommand{\SPIDmdnaShownPTenPct}{4.3}                   % focus_coverage_by_form_item.csv 10-K md_and_a shown_p10
++\providecommand{\SPIDmdnaShownPNinetyPct}{24}                 % shown_p90
++\providecommand{\SPIDmdnaLastShownMedianPct}{16}              % ledger_10k_mdna_last_p50
++\providecommand{\SPIDmdnaShownLtTwentyFivePct}{90.5}          % ledger_10k_mdna_pct_shown_lt_25
++\providecommand{\SPIDmdnaHitsShownMedianPct}{15.6}            % hits_shown_p50
++\providecommand{\SPIDbusinessShownMedianPct}{15.7}            % 10-K business shown_p50
++\providecommand{\SPIDriskShownMedianPct}{10.1}                % 10-K risk_factors shown_p50
++\providecommand{\SPIDtenQmdnaShownMedianPct}{13.2}            % 10-Q md_and_a shown_p50
++\providecommand{\SPIDtenQfinShownMedianPct}{10.6}             % 10-Q financial_statements shown_p50
++\providecommand{\SPIDeightKWholePct}{91}                      % 100 - 8-K pct_over_9000 (8.82)
++\providecommand{\SPIDmsgTruncatedN}{0}                        % stats_keys.json:msg_truncated_14k
++\providecommand{\SPIDmentionsFromLtTwentyFive}{193{,}416}     % stats_keys.json:mentions_from_sections_shown_lt_25
++\providecommand{\SPIDmentionsFromLtTwentyFivePct}{62.7}       % / 308,477
++\providecommand{\SPIDmentionsFromLtTwentyFiveTenK}{84}        % focus_coverage_by_form.csv 10-K pct_mentions_from_lt25
++\providecommand{\SPIDmentionsFromLtTwentyFiveTenQ}{79}        % 10-Q
++\providecommand{\SPIDmentionsFromLtTwentyFiveEightK}{6}       % 8-K (5.74)
++\providecommand{\SPIDplateauLow}{0.4}                         % mentions_vs_length_by_item.csv 10-K md_and_a
++\providecommand{\SPIDplateauHigh}{1.5}
++\providecommand{\SPIDplateauRange}{1.2--1.6}
++\providecommand{\SPIDplateauKchars}{20{,}000}
++\providecommand{\SPIDmentionsPerTenKShownChars}{2.5}
++\providecommand{\SPIDcapexVocabShownMdnaPct}{5.8}             % focus_coverage_by_form_item.csv pct_capex_vocab_shown_over_budget 10-K md_and_a
++\providecommand{\SPIDcapexVocabShownBusinessPct}{10.0}
++\providecommand{\SPIDcapexVocabShownTenQmdnaPct}{8.3}
++\providecommand{\SPIDshortMdnaN}{169}                         % mdna_10k_short_by_year.csv sum(lt30_words)
++\providecommand{\SPIDshortMdnaDenom}{29{,}088}                % sum(mdna_sections)
++\providecommand{\SPIDshortMdnaPct}{0.6}
++\providecommand{\SPIDshortMdnaIncorporatedN}{470}             % sum(lt300_incorporated_by_ref)
++\providecommand{\SPIDrulesFocusRecovered}{84{,}841}           % focus_recall_loss_overall.csv rules_focused
++\providecommand{\SPIDrulesFocusTotal}{324{,}237}              % rules_full
++\providecommand{\SPIDrulesFocusRecoveredPct}{26.2}            % 100 - pct_lost
++\providecommand{\SPIDrulesFocusMdnaPct}{20.8}                 % focus_recall_loss.csv 10-K md_and_a (100 - 79.18)
++\providecommand{\SPIDrulesFocusBusinessPct}{36.9}             % 10-K business
++\providecommand{\SPIDrulesFocusTenQmdnaPct}{25.1}             % 10-Q md_and_a
++\providecommand{\SPIDrulesFocusEightKPct}{84.2}               % 8-K full_body
++
++% ---- rules baseline (results/rules/agreement_overall.csv, agreement_by_form.csv, agreement_by_run.csv, type_agreement.csv) ----
++\providecommand{\SPIDrulesSampleProductive}{30{,}000}         % manifest.json notes.sample
++\providecommand{\SPIDrulesSampleZero}{30{,}000}
++\providecommand{\SPIDrulesGivenLlmPct}{85.7}                  % p_rules_given_llm
++\providecommand{\SPIDrulesGivenLlmTenK}{97.4}
++\providecommand{\SPIDrulesGivenLlmTenQ}{94.4}
++\providecommand{\SPIDrulesGivenLlmEightK}{69.2}
++\providecommand{\SPIDrulesGivenLlmJune}{79.8}
++\providecommand{\SPIDrulesGivenLlmOctober}{86.7}
++\providecommand{\SPIDllmGivenRulesPct}{32.5}                  % p_llm_given_rules_weighted
++\providecommand{\SPIDrulesFlagPct}{39.5}                      % rules_pos_pct_weighted
++\providecommand{\SPIDrulesAgreementPct}{71.2}                 % agreement_pct_w
++\providecommand{\SPIDrulesLlmOnlyPct}{2.1}
++\providecommand{\SPIDrulesOnlyPct}{26.6}
++\providecommand{\SPIDrulesTypeOverlapPct}{84.6}               % type_overlap_pct_codetected
++\providecommand{\SPIDrulesTypeMa}{96.8}                       % type_agreement.csv
++\providecommand{\SPIDrulesTypeDrug}{92.5}
++\providecommand{\SPIDrulesTypeCost}{76.9}
++\providecommand{\SPIDrulesTypeCapex}{55.2}
++\providecommand{\SPIDrulesTypePlant}{40.7}
++\providecommand{\SPIDrulesTypeGeo}{36.5}
++
++% ---- financing / buy-back false-positive classes (results/fp_classes/summary.json, financing_regex_by_type.csv, _by_item.csv, _terms.csv, table5_restated_excl_financing.csv) ----
++\providecommand{\SPIDfinMentions}{22{,}412}                   % summary.json:mentions_matched[0]
++\providecommand{\SPIDfinMentionsPct}{7.3}                     % 7.28
++\providecommand{\SPIDfinAmountB}{17.8}                        % matched_amount_gusd[0] / 1000
++\providecommand{\SPIDfinAmountDenomB}{104.2}                  % matched_amount_gusd[1] / 1000
++\providecommand{\SPIDfinUnregN}{5{,}179}                      % unregistered_sales_matched[0]
++\providecommand{\SPIDfinUnregDenom}{9{,}506}
++\providecommand{\SPIDfinUnregPct}{54.5}
++\providecommand{\SPIDfinEightKPct}{19.2}                      % financing_regex_by_item.csv 8-K full_body
++\providecommand{\SPIDfinFinStatementsPct}{2.4}                % 10-Q financial_statements
++\providecommand{\SPIDfinControlsPct}{0.2}                     % 10-Q controls_and_procedures
++\providecommand{\SPIDfinCapexPct}{52.6}                       % financing_regex_by_type.csv capex_program
++\providecommand{\SPIDfinCostPct}{16.4}                        % cost_reduction
++\providecommand{\SPIDfinMaPct}{7.0}                           % ma_integration
++\providecommand{\SPIDcapexProjectsExcl}{4{,}725}              % table5_restated_excl_financing.csv capex_program
++\providecommand{\SPIDcapexShareExcl}{4.3}
++\providecommand{\SPIDcostProjectsExcl}{5{,}403}
++\providecommand{\SPIDmaProjectsExcl}{37{,}527}
++\providecommand{\SPIDfinProjectsRep}{12{,}027}                % summary.json:projects_matched_rep[0]
++\providecommand{\SPIDfinProjectsRepPct}{9.9}
++\providecommand{\SPIDfinProjectsAny}{13{,}139}                % projects_matched_rep[1]
++\providecommand{\SPIDitemTwoPairs}{77{,}869}                  % summary.json:item2_collision pairs
++\providecommand{\SPIDunregMedianWords}{602}
++\providecommand{\SPIDtenQmdnaMedianWords}{7{,}573}
++\providecommand{\SPIDunregGtThreeK}{1{,}001}
++\providecommand{\SPIDunregGtThreeKPct}{1.3}
++\providecommand{\SPIDswapCandidates}{296}
++\providecommand{\SPIDunregGtThreeKMentions}{425}
++\providecommand{\SPIDfinTermRepurchase}{10{,}886}             % financing_regex_terms.csv
++\providecommand{\SPIDfinTermSeniorNotes}{3{,}375}
++\providecommand{\SPIDfinTermTermLoan}{2{,}263}
++
++% ---- 8-K item-code consistency (results/benchmark_8k/summary.json, item_vs_type.csv scope=all_8k_sections) ----
++\providecommand{\SPIDeightKItemCodePct}{99.7}                 % summary.json:coverage_pct 99.74
++\providecommand{\SPIDbenchMaGivenTwoOhOne}{0.911}             % p_type_given_item ma_integration | Item 2.01
++\providecommand{\SPIDbenchMaGivenTwoOhOneCiLow}{0.903}
++\providecommand{\SPIDbenchMaGivenTwoOhOneCiHigh}{0.918}
++\providecommand{\SPIDbenchTwoOhOneN}{5{,}456}                 % n_item
++\providecommand{\SPIDbenchMaGivenEightOhOne}{0.199}           % Item 8.01 (control)
++\providecommand{\SPIDbenchMaGivenFiveOhTwo}{0.031}            % Item 5.02 (control)
++\providecommand{\SPIDbenchMaBase}{0.095}                      % base_rate_type
++\providecommand{\SPIDbenchItemsGivenMa}{0.467}                % p_item_given_type, Item 2.01 or 1.01
++\providecommand{\SPIDbenchCostGivenTwoOhFive}{0.709}          % cost_reduction | Item 2.05
++\providecommand{\SPIDbenchCostGivenTwoOhFiveCiLow}{0.687}
++\providecommand{\SPIDbenchCostGivenTwoOhFiveCiHigh}{0.730}
++\providecommand{\SPIDbenchTwoOhFiveN}{1{,}700}
++\providecommand{\SPIDbenchCostLift}{92}                       % lift_type_given_item 92.05
++\providecommand{\SPIDbenchTwoOhFiveGivenCost}{0.392}
++\providecommand{\SPIDbenchPartnershipGivenOneOhOne}{0.040}
++\providecommand{\SPIDbenchPartnershipBase}{0.011}
++
++% ---- annotation frame (results/annotation_frame/design.json) ----
++\providecommand{\SPIDannSeed}{20261105}
++\providecommand{\SPIDannStrata}{192}
++\providecommand{\SPIDannNProductive}{1{,}677}
++\providecommand{\SPIDannNZero}{385}
++\providecommand{\SPIDannZeroGatePass}{285}
++\providecommand{\SPIDannZeroGateReject}{100}
++\providecommand{\SPIDannCore}{510}
++\providecommand{\SPIDannDev}{1{,}058}
++\providecommand{\SPIDannHoldout}{1{,}004}
++\providecommand{\SPIDannDeffProductive}{3.4}
++\providecommand{\SPIDannDeffZero}{1.2}
++\providecommand{\SPIDannHalfWidthPrecision}{4.4}              % half_width_95_precision_p0.5 (pp)
++\providecommand{\SPIDannHalfWidthFn}{2.4}                     % half_width_95_zero_fn_rate 0.05 (pp)
++\providecommand{\SPIDannMde}{7.8}                             % mde_gate_recall_diff_at_p0.05 (pp)
++\providecommand{\SPIDannFloor}{10}
++\providecommand{\SPIDannOversample}{2}
++\providecommand{\SPIDannPopProductive}{168{,}230}
++\providecommand{\SPIDannPopZero}{954{,}868}
++
++% ---- time series (results/timeseries/summary.json, first_seen_fixed_cohort.csv) ----
++\providecommand{\SPIDfirstSeenLastYear}{6{,}694}              % all_issuers 2026 first_seen_current_key
++\providecommand{\SPIDfixedCohortIssuers}{1{,}389}             % summary.json:cohort_issuers
++\providecommand{\SPIDfixedCohortMentionsPerIssuerFirst}{9.6}  % 2011
++\providecommand{\SPIDfixedCohortMentionsPerIssuerLast}{8.8}   % 2025 (8.75)
++\providecommand{\SPIDfixedCohortNewPerIssuerFirst}{4.7}       % 2011
++\providecommand{\SPIDfixedCohortNewPerIssuerLast}{2.8}        % 2025 (2.83)
++\providecommand{\SPIDfixedCohortFirstSeenFirst}{6{,}533}      % 2011
++\providecommand{\SPIDfixedCohortFirstSeenLast}{3{,}926}       % 2025
++\providecommand{\SPIDepisodesKthree}{131{,}823}               % summary.json:episodes_k3_total
++\providecommand{\SPIDepisodesKfive}{126{,}318}
++
++% ---- universe and corpus artefacts (results/universe/universe_artefacts.json, results/russell_coverage.json, data/stats.json) ----
++\providecommand{\SPIDdualClassCiks}{18}                       % 2,573 - 2,555
++\providecommand{\SPIDuniverseBoth}{491}                       % implemented_changes.md N10 (universe.json)
++\providecommand{\SPIDuniverseSpOnly}{9}
++\providecommand{\SPIDuniverseRuOnly}{2{,}064}
++\providecommand{\SPIDparseErrors}{14}                         % filings_status_missing.csv parse_status = error
++\providecommand{\SPIDannualPanelFirms}{2{,}499}               % stats.json:annual_panel_firms
++\providecommand{\SPIDrussellGapMax}{441}                      % 2,996 - 2,555
++\providecommand{\SPIDrussellGapMaxPct}{14.7}
++\providecommand{\SPIDfactsheetDate}{September~30, 2026}       % russell_coverage.json:index_as_of
++\providecommand{\SPIDiwvHoldingsDate}{October~23, 2026}       % stats.json:universe.russell_holdings_as_of
++\providecommand{\SPIDiwvFetchDate}{October~27, 2026}          % universe_artefacts.json:russell_scalars.fetched_at
++\providecommand{\SPIDiwvPageDate}{November~4, 2026}           % russell_coverage.json:fund_holdings_as_of
++\providecommand{\SPIDiwvProspectusDate}{July~31, 2026}        % russell_coverage.json:replication_source_title
++\providecommand{\SPIDuniverseJsonSha}{3967b29b}               % universe_artefacts.json sha256 prefix
++\providecommand{\SPIDrussellJsonSha}{758cbebf}
++\providecommand{\SPIDspJsonSha}{50840752}
++\providecommand{\SPIDfactsheetSha}{281e9264}                  % russell_coverage.json:source_sha256 prefix
++\providecommand{\SPIDmethodologySha}{26d2d051}                % methodology_document.sha256 prefix
++\providecommand{\SPIDissuersWithoutFilings}{65}               % stats.json / Table 1
++\providecommand{\SPIDissuersFilingsNoProject}{7}              % 2,499 - 2,492
++\providecommand{\SPIDspWikipediaRows}{503}                    % universe_artefacts.json:sp500_scalars.count
++\providecommand{\SPIDspWikipediaDate}{May~25, 2026}           % sp500_scalars.fetched_at
++\providecommand{\SPIDtickerMapSize}{10{,}430}                 % cik_mapping_scalars.count
++
++% ---- run facts quoted in the text (docs/research_log.md, environment.md, reproducibility) ----
++\providecommand{\SPIDtotalCostUsd}{775}                       % research_log.md 2026-10-27/28 (author-reported total)
++\providecommand{\SPIDworkers}{48}                             % log_summary.json:october.workers_values
++\providecommand{\SPIDrequirementsLockPackages}{88}            % requirements.lock (N12)
++\providecommand{\SPIDpythonVersion}{3.12.15}                  % .python-version (N12)
++\providecommand{\SPIDduckdbVersion}{2.0.0}                    % environment.md (N12)
++\providecommand{\SPIDreferencesN}{42}                         % references.bib entries (L4)
++\providecommand{\SPIDdoiVerifiedN}{36}                        % L4: DOIs re-resolved 2026-11-05
++\providecommand{\SPIDtestsN}{22}                              % tests/test_tables.py (L6)
++\providecommand{\SPIDsnapshotFilesN}{nine}                    % code/audit/spid_snapshot/SHA256SUMS
++\providecommand{\SPIDjuneRunDate}{June~18, 2026}              % log_summary.json:june.first/last
++\providecommand{\SPIDbackfillTimestamp}{October~27, 2026, 13:12}  % backfill_block.json:min_extracted_at

spb/firc0001_uqo

Stage E (writing): main.tex v1.1 / November 2026, abstract rewritten to the audited wording (FIRC0001-THEO-18, ADVE-04, ADVE-05, EDIT-07, ADVE-09), PDF metadata, cleveref; provisional \providecommand macros for Stage D audit numbers (sections/numbers_pending.tex)

Theme