Canadian Small-Cap Cohort (BLDP, OTEX) — Linguistic Readings Only
Lede
The evidence layer for this cohort contains zero forensic signals and zero extracted bullets across both filings — BLDP's 40-F dated 2026-03-12 and OTEX's 10-K dated 2025-08-07. The "above the 95th percentile" annotations on every signal row are an artifact of single-ticker baselines and carry no diagnostic weight. What remains is raw linguistic measurement on two filings, uncorroborated by any disclosed event in the extraction layer. Read accordingly.
Findings
1. BLDP narrative sections run heavy on hedging and forward tilt, on their own terms
In the AIF overview, BLDP registers [BLDP Annual Information Form Hedging language = 7.124], [BLDP Annual Information Form Forward-looking framing = 13.468], [BLDP Annual Information Form Passive voice ratio = 1.425], and [BLDP Annual Information Form Disclaimer density = 1.781]. The MD&A overview is consistent: [BLDP Management's Discussion & Analysis Forward-looking framing = 14.322] and [BLDP Management's Discussion & Analysis Passive voice ratio = 1.746]. These are absolute readings. No peer distribution exists in this evidence set to call them outliers.
2. OTEX hedging concentrates in Item 1A and Item 5
Risk factors carry the highest tentative-marker density observed in the cohort: [OTEX Risk Factors Hedging language = 11.675], alongside [OTEX Risk Factors Forward-looking framing = 9.109] and [OTEX Risk Factors Disclaimer density = 0.513]. Item 5 (market for registrant's equity) shows the highest forward-tilt reading in the file: [OTEX Market for Registrant's Common Equity Forward-looking framing = 15.148], with [OTEX Market for Registrant's Common Equity Disclaimer density = 0.757] and [OTEX Market for Registrant's Common Equity Passive voice ratio = 1.363]. Standard placement for hedging concentration; the levels are what they are.
3. OTEX MD&A is the only section in either filing with a non-zero callback density
[OTEX MD&A Boilerplate callbacks to prior periods = 0.967], paired with [OTEX MD&A Disclaimer density = 0.765] and [OTEX MD&A Passive voice ratio = 0.942]. Callback density at this level in MD&A indicates repeated reference back to prior-period framing. A single observation cannot distinguish stylistic habit from substantive re-anchoring. Flag for the next filing.
4. Neither filing produced a single extracted bullet
[BLDP filing_id=38] and [OTEX filing_id=236] both passed through extraction with no taxonomy hits — no going-concern, no material weakness, no covenant disclosure, no restatement, no executive departure. A 40-F and a 10-K of this size returning empty is the most material observation in this report. It speaks to either uneventful disclosure or extraction-layer coverage. Both possibilities are on the table.
Forensic note
The "above the 95th percentile" labels stamped on every row of the cohort signal summary are mechanical: when the baseline universe holds one ticker, that ticker occupies every percentile at once. The percentile annotations should be ignored. The underlying readings are real — [OTEX Risk Factors Hedging language = 11.675] and [BLDP Annual Information Form Disclaimer density = 1.781] are the two figures worth carrying forward — but they acquire meaning only against a second observation or a populated peer set. Today they are baseline measurements, nothing more.
What to watch
Three things, all measurable against the same evidence stream. First, whether [OTEX MD&A Boilerplate callbacks to prior periods] persists at or above 0.967 in the next 10-K — a repeated callback pattern across consecutive MD&As is the configuration that becomes interpretable. Second, whether [BLDP Management's Discussion & Analysis Hedging language = 2.576] and [BLDP Management's Discussion & Analysis Passive voice ratio = 1.746] drift once a second filing exists to anchor the trajectory. Third, whether bullet extraction returns substantive items from either issuer next cycle. Two consecutive empty extractions on filings of this size would warrant scrutiny of the extraction layer itself, not the issuers.