Built the way a desk builds one: on-chain flow, market structure, momentum, order-book liquidity - then the market grades it, candle by candle.
A setup is Entry / Stop / Targets - assembled the way a trading desk does it, not the way a chatbot does. Before placing a single level the agent reads the whole board: on-chain flow (is smart money accumulating or distributing, what are market makers moving), market structure (the global and the local trend, the levels price actually respected), momentum (RSI, MACD and the rest of the standard toolkit), live order-book liquidity (where real bids and asks stack) and the current smart-money signal picture. The entry lands where those factors agree - and that is rarely wherever price happens to be right now.
Ask for a scalp or a fast trade and it flips to an at-market entry. If your trained agent's style is scalping, that overrides the default - your playbook always wins.
A setup is analytical structure, never a buy or sell instruction.
Where the full analysis says the edge is. The agent fuses on-chain flow, market structure, momentum, order-book liquidity and the signal picture, and places the entry at the level where they agree - not at whatever the current price happens to be. A scalp request, or a scalper's trained style, switches it to an at-market entry.
No. A setup is analytical structure - entry, stop and targets to study - not a recommendation to buy or sell.
Against real price candles: a win when price reaches a target at least 1% from entry, a loss on a stop hit first. Wins and losses both stay on the agent's public record.
Cexlens tracks every large transfer across 20+ exchanges in real time. Spot inflows, outflows and smart-money moves before the market reacts.