That is the exact failure the machine is built to catch. Ideas are searched on —; everything from — on is held back and scored separately. To survive, an idea has to keep making money in those held-back years after trading costs (5 basis points a trade) and beat simply holding the same stock over the same days — so a rising market or lucky timing doesn’t count. Every idea ever tried stays on the books, so the number of attempts is known, not hidden. Result: — are thrown out.

They will. The whole library is re-tested every night on the newest data, and every idea’s status is recalculated: one whose evidence weakens loses its standing, and settled losers are retired from testing (— so far). Confidence counts recent years more than old ones, and an idea that has been fading over the last three years is capped low however good its history looks.

Whalur is a research publisher: it publishes research and signals to subscribers. It never holds, moves or trades anyone’s money — the house wallets are paper only, and members’ wallets are paper too. We plan a review by a securities lawyer before launch.

Stealth by choice. The product, plans and billing are built and working, but we have deliberately not launched. The plan is to close this round, let the paper record run longer, then launch.

Not any single idea — ideas leak. What’s defensible is the machine: the assembled data, the testing discipline that keeps only the — of what it writes that holds up, and a library and paper record that grow every night it runs. A competitor starting today starts with none of that.