53%, until you move the line
A popular belief: “when small traders pile into bets on a rise, the price falls next.”
The data: a basket of coins, and the share of small traders betting on a rise. When that share reached an unusually high point, the price dropped first 53% of the time (60 cases).
Now move the line a little. Call it “unusually high” a bit sooner, and it is 45% (75 cases). A bit later: 46% (46 cases).
A result that changes when you nudge the line is about the line, not the market.
If a number only works at one setting, would you trust it?
