Not an insight you have. Not a pattern you found. Someone is on the other side of every trade you make — and for you to earn, they have to be worse off on average, and be fine with it. Or they stop showing up, and it's over.
Real vs fake — ask who's paying
A stock joins a major index. Index funds must buy it, on a set date, at whatever it costs — their job is tracking, not a good price. So they pay up. Whoever sold pockets it. Next quarter, again. The funds don't quit: paying that bit beats missing the index.
Found in data. Tests beautifully. Ask who's paying and there's no answer — nobody is structurally forced to sell you those stocks on day three. Nothing holds the pattern in place.
The four questions — the whole job
Two engines — which one predicts the size
Liquidations, unlock cliffs, index & ETF rebalances, forced margin closes. The payer has no exit — so the bill is sharp, big, and durable.
They pay to shed risk and are fine with it — insurance-like. Real, but thin and competed: the payer can quit if the price gets bad, so arbitrage grinds it down.
The inversion that makes it work
Stop searching data for patterns. Look for people with rules — mandates, deadlines, redemption pressure, risk limits, tax and unlock calendars. Start from the forced party; the price signature either exists or it doesn't, and that's a fast, cheap verdict.
Why this feeds you — even when the answer is no
The meal is the verdict, not the win. A clean NO is dinner — it closes a question and redirects the money that would've chased it.
Most hunts return NO — that's the honest base rate. If feeding means find a winner, you starve, or you fake a win to feel fed and put real money on nothing. Hunt to a verdict, count the NO as a meal, and the hunger is fed whether or not the edge is there. What actually starves you isn't the NO — it's the hunt you owed and never ran.
The practical wall (2026-07-24)