As of September 5, 2026, the lab had produced 174 immutable evaluation reports. Seven were complete. The other 167 were blocked: 97 did not contain enough detail to test honestly, and 70 were not supported by the evidence attached to them.
That does not mean 167 brilliant strategies lost money. It means most internet “strategies” are a chart, three indicator names, and a sentence like “wait for confirmation.” A computer cannot backtest a shrug.
The smaller useful pile
A bounded research run made five passes through the material and kept 22 claims worth more work. From there we ran hundreds of direction tests, short-side regime tests, and nearby-parameter checks across multiple coins and timeframes.
The tester uses the next bar after a signal, includes fees and slippage, keeps time in order, and checks more than win rate. Expectancy, average win, average loss, payoff ratio, profit factor, drawdown, trade count, and cost drag all matter. A strategy winning 80 percent of tiny trades can still get murdered by one loss.
Uncertain is a real answer
A promising result does not have to be thrown away because it lacks enough trades. It gets marked uncertain and can be tested later under a versioned rule set. It just cannot be advertised as proven.
One short-side candidate earned a limited simulated trial after positive walk-forward and stress results, but its historical sample was still below the normal promotion requirement. That is exactly the kind of thing the paper desk is for. This is simulated research, not a prediction of future profit.