Canadian Energy Cohort — Six 40-F / 10-K Filings, 2017–2026
Lede
No event-level disclosures were extracted across the six-filer cohort — no covenant items, no auditor changes, no going-concern language. What remains is a linguistics-only read, and the read is a clean cohort split: pipeline operators write with high certainty and thin hedging; integrated producers write with elevated tentative markers, future-tilt, and passive construction. The frame is the spread between [Annual Information Form Certainty / hedge imbalance = TRP 0.552 above the 95th percentile] and [Annual Information Form Certainty / hedge imbalance = CVE 0.067 bottom 10%] — a 0.485-point gap against a baseline p95–p25 range of 0.209.
Findings
1. Certainty divergence sorts on business model, not on filer idiosyncrasy
TRP anchors the top of the cohort at [Annual Information Form Certainty / hedge imbalance = 0.552 above the 95th percentile]; CVE anchors the bottom at [Annual Information Form Certainty / hedge imbalance = 0.067 bottom 10%]. Cohort mean sits at 0.379. The split tracks pipeline-versus-producer cleanly enough that the language itself is functioning as a business-model tell. This is structural, not anomalous — but it is the cohort's most legible signal.
2. CVE: three independent hedging signals converge in the same section
In the AIF overview, CVE registers [Annual Information Form Hedging language = 3.233 above the 95th percentile], [Annual Information Form Vague attributions = 0.231 above the 95th percentile], and [Annual Information Form Passive voice ratio = 0.947 above the 95th percentile] simultaneously. Three orthogonal hedging measures landing above p95 in one filer's one section is the observation. The certainty deficit noted above is not an isolated reading — it is what that convergence produces.
3. CNQ: forward-tilted and disclaimer-heavy, but actively voiced
CNQ posts [Management's Discussion & Analysis Disclaimer density = 0.772 above the 95th percentile] against a cohort mean of 0.204, and [Annual Information Form Forward-looking framing = 25.910 above the 95th percentile] against a cohort mean of 12.294. Passive voice runs the other way: [Management's Discussion & Analysis Passive voice ratio = 0.964 bottom 10%]. The filer is adding disclaimer scaffolding and forward-looking reach while keeping the syntax in the active voice. That is a mixed profile worth tracking, not interpreting from one filing.
4. ENB's disclaimer load is real but stale
ENB shows [Annual Information Form Disclaimer density = 0.896 above the 95th percentile] against cohort mean 0.179, plus [Management's Discussion & Analysis Hedging language = 2.748 above the 95th percentile] and [Management's Discussion & Analysis Vague attributions = 0.257 above the 95th percentile]. The filing is the [ENB 40-F dated 2017-02-17] — nine years behind the rest of the cohort. The signals are present in the document; they are not contemporaneous, and they should not be read as a current-period read on ENB.
5. IMO is structurally non-comparable in this cohort
IMO is the only 10-K filer in a cohort of 40-Fs. Every item_* subsection in the summary shows IMO as both the low and the high extreme, with baseline p25 = p50 = p75 = p95 — for example, [MD&A Disclaimer density = IMO 15.845 above the 95th percentile] with all percentiles equal to 15.845. These "above the 95th percentile" annotations are artifacts of an n=1 baseline, not signal. IMO's section-level numbers cannot be benchmarked against this cohort.
Forensic note
The buried observation is the certainty gap. [Annual Information Form Certainty / hedge imbalance] spans 0.485 between TRP and CVE; the baseline p95–p25 range is 0.209. Two filers in the same regional cohort sitting on opposite sides of the full baseline distribution means the language is sorting along a structural axis — midstream certainty versus upstream hedging — and the CVE side of that axis is reinforced by [Annual Information Form Hedging language = 3.233], [Annual Information Form Vague attributions = 0.231], and [Annual Information Form Passive voice ratio = 0.947] all clearing p95 in the same section. Three signals, one filer, one section. That is where to look first if event disclosures surface later.
What to watch
Next CVE 40-F: whether [Annual Information Form Hedging language] and [Annual Information Form Vague attributions] retreat toward the cohort mean (1.855 and 0.083 respectively) or entrench above p95. Entrenchment in the absence of a corresponding event disclosure is the read. Next CNQ 40-F: whether [Management's Discussion & Analysis Disclaimer density = 0.772] holds or climbs while [Management's Discussion & Analysis Passive voice ratio = 0.964] stays below p10 — scaffolding without syntactic distance is the pattern to confirm. For ENB, a current-year 40-F is the precondition before any of the 2017 signals are treated as live.