Catalyst edge

This was the one edge we featured as validated. On 2026-06-15, a study of the 2024–2026 SEC record found that when 2 or more company insiders buy their own stock within ~10 days, it appeared to beat the S&P 500 over the next week. Those original numbers, kept here as the historical record:

Why it did not hold up: the original test covered only ~2 years, and insider clusters bunch together in time, so its statistics quietly counted overlapping events as if they were independent. That combination made a noisy pattern look real. On 2026-07-30 we re-ran it the honest way: 18 years of data (2008–2026), 2,934 cluster events, 27 rule variants, with overlap-corrected statistics. The corrected t-stat fell from 2.6 to 0.54 (far below significance), the placebo test itself came back significant (a sign the ticker selection, not the clustering, drove the result), the median trade lost money, the two halves of history disagreed on direction, and 0 of 27 variants passed. Verdict: no tradeable edge. This page exists to show what failed, and that includes our own headline finding.

We only feature edges that survive a real test on real history. These were checked and did not show a tradeable edge:

• Insider cluster buying — our former headline edge; refuted 2026-07-30 on the full 18-year record (full story in the card above).
• 8-K filing drift — buying the morning after corporate filings (incl. earnings) slightly lost to the market across 4,000+ filings; the move is already done by the time you can act.
• ORB 0DTE options day-trade — no durable edge over a full year of real data (the option leg lost).
• Short-squeeze setups & “follow Congress” — survivorship bias and a 45-day reporting lag; ruled out before testing.

How many distinct insiders bought each name. Informational only: our 18-year test found no tradeable edge in clusters. Click a bar to drill in.

Stocks where 2+ different insiders recently bought, shown as an informational feed of who is buying what. Not an edge: our 18-year test (above) found no tradeable advantage in these clusters. Context, not a buy list.