A ruthless person: 90% of people guess whether US stocks will rise or fall, but the direction is a random walk. I changed my thinking: I don't guess the direction, I only test if it can withstand a beating. The result is particularly counterintuitive - it doesn't fall on average under extreme pressure, but the probability of a big drop doubles. Now the buffer is thick, AI crowding is at the 94th percentile, and September is the most ferocious.