Context
Roughly forty-five to fifty thousand “P”-coded Form 4 purchase filings hit EDGAR every year. Almost all of them are noise: routine, small, or mechanical.
The interesting subset is narrow. When a cluster of senior officers — CFO, COO, directors, the people looking at the real numbers — buy their own company’s stock with their own money, above a threshold that means something to them personally, that is a different signal from a scheduled option vesting.
The problem
The signal-service market is crowded with newsletters selling tips, and the default way to sell one is to be vague about the method and loud about the wins. That positioning attracts exactly the audience you do not want: people looking for something to copy rather than something to evaluate.
The site had to do the opposite. State the rule, state the threshold, show the backtest, and be readable by someone who intends to disagree with it.
What I did
The edge is published rather than teased. A $500,000 dollar-size threshold. A 3× role weighting on the numbers-facing executives. Cluster detection over a sliding window. Anyone can read the method and decide it is wrong — which is the correct relationship to have with an audience of independent decision-makers.
The headline claim is a 82.9% calendar-year 2024 backtest, with named worked examples rather than anonymous percentages: ATGE +68%, OSUR +52%, TLS +197%. Naming the tickers is a commitment; a reader can go and check them.
Structurally the site is a thin static surface. It is a delivery channel, and saying so is part of the pitch — the actual product is the screening pipeline behind it, which is why the weekly output can be free. The filtering work that would otherwise be a full-time job is done by the scanner.
The result
A research service that gives away its output and keeps its method visible, aimed at readers who want to assess a thesis rather than receive one.
Worth being straight about the status: the live build still carries an “under development, content is placeholder” notice. The methodology, the threshold and the case studies are real; the copy and the publication cadence are still settling.
