How the trading bot would connect, trade, and stop

The working order of Captay is meant to be dull on purpose. You read the rule, you watch it on paper, you attach a key that cannot withdraw, you write the limits, and only then could a strategy be armed. This website is stuck before the first of those steps that touch an exchange. Nothing in the sequence below is running.

Paper before any ticket

Paper mode, when it exists, would record hypothetical orders against public prices and would not transmit them. It is a way to see whether a schedule, a grid, or a rebalance fires more often than you thought. It is a poor imitation of the live book: it will not show queue position, partial fills, or the moment a price gaps through your level. Treating a paper log as proof of a future result would be a misuse of the tool.

Even that rehearsal is unavailable. There is no market feed attached to these pages and no simulated fill to inspect. The description is here so the order of operations is visible before a cabinet opens, not so you can practice today.

A key that can trade and cannot leave

The connection, later, would be a trading API key created on the exchange by you. The intended permission set is read and trade, with withdrawal and internal transfer switched off. Captay would not ask you to paste a seed phrase, and this website has no field that accepts a key at all.

You would keep the ability to delete the key on the exchange. That deletion is the real off switch for access. The product's own halt only matters while the key is still valid. If a key leaks, rotating it on the exchange matters more than any toggle drawn in a cabinet that does not yet exist.

Rules, then the limits around them

A strategy is a small piece of arithmetic: buy this much on a clock, quote these levels inside a range, or trade when weights drift. Around it sit four constraints that are not part of the strategy's opinion. They are the share of capital it may touch, the largest position it may hold, the drawdown from a high-water mark, and the loss allowed since the session opened.

A pre-trade check would refuse an order that breaks one of those constraints. The strategy does not get a vote. If the check fails, the order is not sent and the journal records the refusal. That is the design. There is no checker running on this site, because there is no order path.

The journal is a log, not a score

The journal is meant to be an ordered list of decisions: armed, refused, limit hit, halted, key added, key removed. It is not a performance report and it will not be decorated with a return figure. A day with no trades is a valid page in that log. A day that cannot be reconstructed is a failure of the product, which is one reason the cabinet stays closed until the log exists.

Stopping is a person-sized act

An emergency stop would cancel resting orders and block new ones. It would not close the position by magic and it would not undo a fill. Closing is itself a trade, with its own price and fee, and it would have to be a separate, explicit act. Automatic restart is refused in the design: after a limit or a halt, a person re-arms. A rule that turns itself back on has already stepped over the line.