Asymmetry screener
What it does
Finds companies where the equity is a thin slice of a large enterprise value, so a small move in the enterprise produces an outsized move in the stock. Both directions. Every row shows the upside multiplier and, beside it, the percentage fall in enterprise value that takes the equity to zero, because they are the same fact and showing only one of them makes the trade look like something it is not.
It is a research tool. No trading logic, no sizing, no recommendations.
The hard part
Filers tag debt differently, and the naive screener gets it wrong in both directions at once. Some companies carry hundreds of millions in convertible notes under a tag the usual query never asks for, so they show zero debt and never appear. Others tag the same borrowing under five overlapping concepts, so a sum double- or triple-counts. Lengthening the tag list fixes one case and worsens the other.
Debt therefore resolves by ladders: prefer a single total tag, fall back to component tags in a fixed order, never mix rungs. Every rule was confirmed against live filings before it became code, and a validation suite replays those filings on every change.
What I learned about tools that produce numbers
- A ranked list with no trigger is a reading list. The screen is paired with a filings monitor and an append-only ledger of what was flagged and what happened next, so the tool keeps score against itself.
- Citations rot. A command re-fetches every cited filing and confirms the quoted text is still there.
- The expensive input is analyst reading time, not compute. That observation became the next project.
Details of the fund, the names screened and the curated registries are not published.