The machine can see the whole market now. It measured whether that’s worth anything.

Over one week the machine went from watching about forty coins to watching, more or less, all of crypto. Three exchange data pipes, a multichain screen, a nightly snapshot of every listed name above a dust floor. Thousands of price series, none of them tradeable, all of them watched. The bet was my thesis: more eyes means more opportunities.

Then we did something I’m prouder of than the widening. We measured whether the eyes are worth anything — before acting like they are.

The test was simple and fixed in advance. Take the machine’s own entry signal — a 20-day breakout — and scan every watched series for the last 90 days. Every breakout, judged by what happened over the next ten days. In the twelve names the machine can actually trade, breakouts followed through 26% of the time. Across the roughly 1,400 series it can only watch: 24%. Same coin. And the median breakout lost money everywhere — worst in the thin stuff, where the typical one was down 10% ten days later.

So the honest reading, which the report’s own rules forced: the watched world is not hiding a better version of our pattern. The eyes are an option on some future market, not income in this one. The machine’s response was to freeze — no new venues, chains, or instruments until the evidence changes — and spend the effort on better tools instead of more coverage.

I wanted more eyes. The machine measured them, told me they’re not alpha, and stopped. That’s the relationship working exactly as designed.

— Adam, Studio Amadeus