Unrealized-PnL wallet ranking, 6 chains

Ranks wallets by total profit on ONE CHAIN over a selectable window. Chain-scoped only: there is no way to narrow this to a single token, and none is offered. ONE-DIRECTIONAL RANKING: highest unrealised gain first, always — there is no losers or reverse-sort direction, and none is offered; the source itself now rejects every value of its own reversal parameter but its own default. DISTINCT FROM realized-pnl-leaders: that endpoint ranks Solana wallets by REALIZED profit since a fixed epoch; THIS endpoint ranks wallets on six chains by a profit figure that is almost always UNREALISED. The two are not fallbacks for each other and never share a field meaning. IMPORTANT: totalPnlUsd is overwhelmingly an UNREALISED, mark-to-market figure on positions still open, not banked profit, on every chain this serves. Measured across two independent full runs of 120 ranked wallets on six chains: only ONE row ever had a majority of its profit realised, and most rows claim more profit than the USD notional the wallet has ever traded — the worst measured row claimed $18,528,065 of profit on $0.12 of volume. Always read realizedPnlUsd (profit already sold — a real 0 means nothing has been sold yet, not missing data) and volumeUsd beside totalPnlUsd before treating a row as a trading record. The cause of the gap is not established: a thin position marked to an illiquid price, and a mis-recorded cost basis, both fit the same data. Only the head of a ranking is stable between requests — the tail can reshuffle within minutes, and how far the instability reaches depends on the chain. A row identifies a wallet by address alone: no name, label or score travels with it. A cache may sit in front of this endpoint, so a ranking already a little stale is a normal answer, not a bug.