Open-Market Insider Purchases Core factor

Our only Tier 1 result to date

Net insider buying (Purchase minus Sale, both open-market transaction codes) tested as a genuine null in our own pipeline - see Raw (Net) Insider Score. This construction isolates purchases only, dropping sales entirely, to test the specific academic hypothesis that insider buying carries a different, stronger signal than net insider activity.

The real result

Tested against 200,522 real factor observations across 2,131 companies, 283 monthly rebalance dates from 2012-12-01 to 2026-09-01 - our largest real evidence base to date. Mean rank correlation is positive at both horizons (60d: +0.017, 252d: +0.019) - the direction the academic literature on insider purchases (Seyhun, 1986; Lakonishok & Lee, 2001) predicts. This is this project's first, and so far only, result to clear every one of our strict Tier 1 gates.

Quality scorecard

Dimension	Score
Sign consistency	85.5%
Signal strength	0.014
Horizon consistency	100% (both horizons agree on sign)
Crisis robustness	100%
OOS stability	68.3%

Why "core factor," and the honest caveats that come with it: all four Tier 1 gates independently cleared - but sign consistency (85.5%) clears the 85% bar by a real, narrow margin, not an overwhelming one. The effect size itself is small (0.014) - "core factor" describes the consistency and robustness of the direction, not the magnitude of the edge. Survivorship bias (today's tracked universe, not a point-in-time historical one) is the same disclosed limitation every result on this page carries.

Not yet queryable via its own endpoint: this construction (compute_insider_purchase_score) currently exists as a validated calculation used by our internal research harness, not yet wired into its own GET /companies/{ticker}/... route. Flagged here honestly rather than documenting an endpoint that doesn't exist yet.

Full validation methodology: FACTOR_VALIDATION_METHODOLOGY.md.

Free early access to the API: Get API access →