Methodology

Data Sources

Stock Score — v2 Model (6 sub-scores)

Each stock is scored across six sub-categories derived from SEC regulatory filings. Sub-scores are each 0–20. The Model Score (0–100) is a profile-weighted composite — not a simple sum of the sub-scores.

ETF Scoring — a Separate, Fund-Specific Model

ETFs are never scored by the company model above — a fund holding hundreds of securities isn't evaluated the same way as an operating business. U.S.-listed ETFs are scored by a dedicated model sourced from official SEC EDGAR fund filings (Form N-PORT, N-CEN, and Risk/Return Summary data), covering four categories: cost, holdings diversification, transparency, and structural risk. Every ETF on this site ends up in exactly one of five explained states, never an unexplained placeholder:

Minimum evidence gate. An ETF only receives a numerical score if it clears all of: weighted supported coverage of at least 50%, real evidence in at least three of the four categories, and at least one of those being a core category (cost, diversification, or structural risk — transparency alone is not sufficient; with only four categories total, clearing three structurally guarantees a core category is included). A missing category is excluded entirely from the calculation, never assigned a placeholder value or scored as zero.

Canadian ETFs.A source-and-licence audit of every major Canadian ETF issuer (BlackRock/iShares Canada, BMO, Vanguard Canada, Global X Canada, Hamilton, Purpose, TD) found that each restricts use of its website content to personal, non-commercial purposes, with no blanket commercial-reuse grant and no Canadian regulatory open-data equivalent to SEC EDGAR's bulk fund filings. As a result, Canadian ETFs currently carry identity and structure description only (issuer, index/strategy tracked, asset class, geographic exposure) — never a reproduced management fee, net assets, or holdings figure, and never a numerical score — and are classified Score Pending until a commercially displayable fund-specific source is identified.

Score Pending, Special Structure, and Historical ETFs are never mixed into the Full or Low-Data ETF Rankings — each appears in its own, clearly separated section. See the Full & Low-Data ETF Rankings → · See the ETFs with Scores Pending directory → · See the Special-Structure Products directory → · See the Historical & Closed ETFs directory →

Low-Data Score & Score Pending

A group of Canadian companies without SEC EDGAR filing coverage are evaluated by a separate, less reliable model sourced entirely from free, explicitly commercially-reusable government open data (e.g. OSFI's Open Government Licence datasets, GLEIF, the Canada Energy Regulator) — never from SEC filings, a commercial data vendor, or the company's own website. Every such company ends up in exactly one of two states: a numerical Low-Data Score, or a non-numerical Score Pendingnotice. A company is never penalized with an artificially low score just because most of its data is missing — if there isn't enough evidence to say anything meaningful, its score shows as Pending instead of being forced.

Minimum evidence gate. A company only receives a Low-Data Score if it clears all of: weighted supported coverage of at least 35%, real evidence in at least two of the five financial categories (growth, profitability, financial health, momentum, dividend — theme exposure never counts toward this), and at least one of those two categories being a core category (growth, profitability, or financial health — momentum and dividend alone are not sufficient). Evidence must come from an approved government source with a resolved entity mapping, and any subsidiary-only evidence is attenuated toward neutral rather than used at face value. Critically, subsidiary-only (partial) evidence also contributes only a fractionof its category's weight toward the 35% coverage requirement — a 20-point category supported by a subsidiary representing 60% of the parent's operations contributes 12 points of coverage, not the full 20, since counting it at full weight would overstate how much of the parent company is actually known. A company that fails this gate shows Score Pending — Limited Data: no score, no percentage, no positive or negative label of any kind — just a clear explanation that we do not currently have enough commercially displayable company-specific financial data to calculate a reliable score, which is not a negative assessment of the company.

For companies that pass the gate, the model uses the same six category weights as the full company model (Growth 15%, Profitability 20%, Financial Health 20%, Momentum 20%, Dividend 15%, Theme 10%), but only categories with real, verified (or properly attenuated, entity-limited) data participate — a missing category is excluded entirely, never assigned a placeholder value or scored as zero. Available category weights are re-normalized before averaging, then a weighted data-coverage penalty is applied and shown separately, never hidden inside the final number: −2 points at 80%+ coverage, −4 at 65–79.99%, −6 at 50–64.99%, −8 at 35–49.99%. A confidence label follows the same tiers (Moderate confidence — incomplete data / Limited confidence / Low confidence / Very low confidence). Every Low-Data Score stays labeled "Low-Data" even at high coverage — it is a different source architecture than a full SEC-based score and the two are never presented as equivalent.

Neither a Low-Data Score nor a Score Pending status is ever mixed into the primary company rankings — each appears in its own, clearly separated section. See the Low-Data Company Rankings → · See the Canadian Companies with Scores Pending directory →

Historical & Acquired Companies

A company that is no longer independently traded under its listed ticker — because it was acquired, taken private, or delisted — is never scored and never appears in current-company discovery: not the Research directory, not the Screener, not the Low-Data Company Rankings, and not the Score Pending directory. Instead it receives historical treatment: a clear statement that the security is no longer independently investable, the effective acquisition/delisting date, and the acquiring or successor entity, all sourced from official acquirer or regulator announcements. Historical pages remain reachable by direct search (marked with a distinct "Historical" badge) and are listed in their own, separate directory. See the Historical & Acquired Canadian Companies directory →

Risk Classification (6 bands)

For v2 stocks, risk is assessed using four filing-derived factors: leverage (0–4 pts), interest coverage (0–3 pts), earnings volatility — downside semi-deviation over 8–10 quarters (0–4 pts), and revenue size (0–3 pts). A higher composite score means stronger fundamentals and lower risk.

Goal Profiles — Weight Distribution

The goal profile determines how the six sub-score weights are distributed to compute the Model Score. The Scanner scores each stock under all six profiles simultaneously.

Theme Score (8 factors)

Each macro theme is scored on the strength and investability of the underlying structural trend. The theme score does not reflect individual stock quality — it assesses the tailwind itself.

Precomputed Architecture

Running a full stock analysis — fetching SEC EDGAR filing data, scoring all sub-categories, generating AI-written analysis, building the risk profile, and running theme scoring — takes time and API cost per request. Rather than doing this live for every visitor, the Python pipeline runs on demand and writes all results to static JSON files. This website reads those files at build time and generates static HTML.

The data you see is a snapshot as of the last update date shown on each report. Scores, metrics, and AI analysis do not update in real time. Always check the “Updated” date and verify key data through primary sources (company filings, official stock exchanges).

Limitations

• Scores are generated by an algorithm. They do not reflect qualitative factors a human analyst might weight differently.
• Financial data is sourced from SEC EDGAR regulatory filings. Errors or omissions in filed data propagate through the model. Not all financial metrics are disclosed in the same format by every filer.
• The model uses fixed formulas. Different sector dynamics, extraordinary items, or unusual capital structures can produce misleading scores for specific companies.
• A high score does not mean the stock is suitable for you. It means the model found the data profile attractive under the given goal profile.
• Coverage is limited to the stocks in the research universe. Absence from this site does not imply anything about a stock's quality.
• AI-generated analysis may contain errors, outdated characterisations, or hallucinations. Treat it as a starting point for research, not a conclusion.
• A subset of Canadian-only tickers and ETFs are not yet covered by the v2 pipeline. Their pages show basic identification and theme information while expanded research is being prepared.