SC Alpha — methodology v1

ONE RULE, FROZEN, TIMESTAMPED.

Alpha is a rules-based seed selection method built on public SEC Form D filings and a proprietary roster. Six parts, each measured: the data (650,101 filings, 2013–2026, rebuilt from public files with one command); the roster (403 partners at 101 firms with dated tenure, forward-only); the gates (new Form D, equity, issuer under five years, at most $10M, roster partner as director not officer); the layers on top (an above-cap band, multi-partner and amendment tilts, each published with its n); the network (every print's people resolved and ranked for a warm path, and the list of firms whose seeds are open at filing); and the record (a daily out-of-sample log, timestamped, 45-day public delay). The filing is public. The roster, the layers, the network and the record are not.

THE BACKTEST

Corpus: 650,101 Form D filings, 2013 through mid-2026, from the SEC's published structured datasets. Outcome: the same company files a $15M+ institutional round within 36 months. That label undercounts true success, so rates are floors and lift is the robust statistic. At higher bars the edge holds: prints reach $100M+ follow-ons at 9.6x the base rate.

CohortnGraduatedLift
All qualifying seed filings (base)80,1953.09%1.0x
Alpha prints49428.34%9.2x

WHAT DIDN'T WORK

A backtest you can trust reports its kills. These hypotheses failed, and we publish them:

Pre-raise filings (no sale yet)0.1x
Placement-agent-sold deals0.4x
General-solicitation (506c) deals0.3x
Disclosed zero revenue0.2x

All inputs are public SEC filings, requested with a declared user agent well inside SEC rate guidelines. Internal research, not investment advice.