Coverage is a registry, not a folder of PDFs: every name carries pillars with an evidence tier, the conditions that would break the thesis, dated catalysts, the red-team's dissent and fixed-horizon outcomes. Statuses say covered, never whether a name is held. Click a row to open the file.

Each workbook stitches four to five fiscal years of income statement, balance sheet and cash flow from the company's own 10-K XBRL on SEC EDGAR, projects from the trailing twelve months of the latest 10-Qs, and runs a ten-year FCFF DCF with a stated WACC and a WACC × growth sensitivity, checking its own revenue, net income and share count against the SEC companyfacts API before it is trusted. The explorer below recomputes the same arithmetic in your browser and reports the growth the market price implies — bend the assumptions and watch both move. It is a transparent scaffold, not a target price.

Every item is a verdict I actually wrote down, with the date, the numbers that carried it, and the evidence tier of each number. Kills and passes are kept on purpose — a ledger that only shows winners is marketing.

Computed by my own producers from primary feeds — CBOE option chains and VX settlements, CFTC positioning, FINRA short files, OCC volumes, FRED, the US Treasury, the RBA, the OFR, the New York and Cleveland Feds, Binance funding, Tiingo closes — and rebuilt on every refresh. Every tile carries its as-of date; anything past its threshold turns STALE rather than lingering.

Dealer positioning by name — where hedging flows amplify or damp a move into earnings. Computed here from CBOE's own delayed option chains: net dealer gamma per 1% move, the gamma flip, the call and put walls, the straddle's implied move and front ATM vol, for ten coverage-weighted large caps and the SPY complex. Nothing on this page says what is held; the name set is fixed in the build script, not drawn from any book.