Placebo replays every trader's own trades at random times. Same coins, same direction, same number of decisions. If the real record can't beat its own placebo, the profit came from the market's direction, not from skill.
Every opening fill, summed per coin per hour, is one decision. We score its 24-hour forward return.
The whole timeline is shifted to 1,000 random offsets. Coins, direction, count and spacing stay; only when changes.
The p-value is the share of clones that did as well. Benjamini–Hochberg at q = 0.10 across every address tested.