Methodology

How reports are structured

Every report is one of four types: primer (industry or business-model groundwork, no rating), initiation (first rating and price target on a name), update (revises a prior report — the prior report is never edited; an update is a new dated entry that links back to what it revises), or portfolio-note (cross-coverage commentary). Each rated report states a rating (Buy / Hold / Sell), a price target with its valuation basis, the price and benchmark level on the publication date, and a per-report disclosure block.

Valuation approach

[EDIT BEFORE LAUNCH: describe your actual valuation approach — e.g., which names get a DCF vs. multiples, how you set exit multiples, how you treat Korean holdco discounts. Placeholder text must not ship.]

Scorecard conventions

• All returns are price returns, excluding dividends, in the security’s local currency (USD or KRW).
• Excess return = absolute return − benchmark return over the same period (arithmetic difference). Each report names its benchmark (e.g., SPY for US names, KOSPI200 for Korean names) and records the benchmark level on the publication date.
• A win is directional: a Buy wins if excess return is positive, a Sell wins if it is negative. Holds are excluded from the hit rate.
• Aggregates (hit rate, average excess return) are equal-weighted across calls, mixing local-currency returns without conversion.
• Closed calls (including thesis-broken) are frozen at their closing price and closing benchmark level; the current price snapshot never restates them.
• Superseded reports keep their own scorecard row, measured from their own publication date at their own rating. The coverage table shows only the latest report per name.
• If a covered name splits, the entry price in the affected report’s metadata is retroactively split-adjusted with a visible footnote in that report.
• Prices are updated manually (target cadence: weekly) and stamped with an as-of date shown on every page that uses them. Every number on the scorecard is recomputed from published report metadata plus that snapshot at build time — there is no hand-edited scorecard.

Data sources and tools

Research draws on public filings (SEC EDGAR for US names; DART for Korean names) and company disclosures, read directly. Where terminal or vendor tools (e.g., Bloomberg, FactSet) are used, they are used manually as research aids — they are not data sources of this site. Site prices are recorded manually from public quote data and stamped with their as-of date.

The FaustCalc risk lens

FaustCalc is a risk-analytics engine I designed and built (Spring Boot, Python, Postgres, deployed on Railway). It computes per-ticker risk features — downside/crash behavior relative to market benchmarks, style and quality characteristics, and filing-sentiment analysis — and a personal risk-capacity equation. Where its per-ticker features inform a report’s risk section, the report says so. A formal, per-report FaustCalc risk appendix is in progress and will be added to new reports when it ships; nothing on this site is auto-generated by FaustCalc today.