Congressional conviction

One legislator is noise. Several, in the same weeks, is not.

Sitting members of the House and Senate disclose their trades under the STOCK Act. Most of what they hold is what any diversified portfolio holds — that is the base rate, not a signal. This page ranks companies by how far they depart from it: how many independent legislators bought, how tightly clustered in time, and how one-sided.

How this is calculated, and what it cannot tell you

The rank combines five factors: the number of distinct legislators buying (logarithmically — the second buyer confirms, the tenth just tracks the index), how tightly those purchases cluster in time, the ratio of buys to sells, the estimated dollar volume, and recency. ETFs, funds and bonds are excluded, as are exchanges, which are neither a purchase nor a sale.

It does not weight by company size. A purchase in a small company should count for more than the same purchase in a mega-cap, and this score does not yet make that distinction — so it under-rewards exactly the companies where agreement is most meaningful. That is a known limitation, not a modelling choice.

These filings are not timely. The disclosure lag column is the average number of days between a transaction and its publication. It is routinely measured in weeks. Nothing here can be acted on as news, and it is not meant to be.

The company columns come from a different window. Insider figures, market capitalisation and headcount describe the companies Quorum tracks — a universe built from the last 30 days of insider filings and the last 90 of congressional ones, and capped at the 50 highest-conviction congressional tickers. This page ranks a longer window, so a company can sit high here and still show nothing in those columns. That gap is coverage rather than an error.

Amounts are disclosed as ranges, and we use the midpoint. Every dollar figure taken from a congressional filing is an estimate — including the purchase total shown under the buy count, which is why it is labelled est. and rounded. A member who reports “$15,001 – $50,000” is counted at $32,500, and nothing in the record says where in that range the truth sits. There is no sale total beside it, and that is an absence rather than a zero: the daily scoring run sends an estimated purchase volume and no equivalent for sales, so publishing one would mean inventing it.