Reactor for the active trader who wants to improve

You can start wherever you need help: use the chat for analysis, set up an Agent with a specific role, or connect several Agents into a workflow. Your instructions decide the way of analyzing, the room for judgment and the authority to execute.

You can ask Reactor to test the idea against market data: trend across several timeframes, momentum, volatility and other information available for that market.

You can set the approach you care about — for example, look for confirmation of a breakout, or check whether the entry is too late. You can also compare the analysis of several models on the same opportunity.

You can describe what you're looking for and use Reactor to scan the market. For example: assets that fell within a certain range, a defined price pattern, or a combination of conditions.

The results become a list you can keep analyzing. When the search fits your need, you can save the instructions and run them again, or pass the results to another Agent.

Reactor lets you break the research into stages. You start by defining the opportunity and collecting its occurrences. With the candle tools you can replay price movement after each occurrence and check how far it moved in the trade's favor, how far against it, and how long it took.

That way you can research the opportunity before adding an Agent that picks which trades to take. Later you test the quality of the decisions and the trade management separately.

In a real example, we scanned for a Bullish Engulfing pattern every four hours. We researched the movement after the signals and, based on the findings, chose a 3% stop-loss and a 4.5% profit target. Then we ran a Buyer that buys the signals.

As of preparing this example, by the measurement provided, 800 trades closed with a 65% win rate. In this case the Buyer doesn't filter with judgment; that lets us collect results before testing an extra decision layer.

You can start from the trade report and look at the overall picture: profit and loss, win rate, average win versus average loss, and how trades were closed.

When there's also a record of the signals and the Agent runs, you can connect the opportunity, the decision, the entry and the management. The candle tool lets you check what happened to price afterwards.

That way you can test questions like: was the entry late? Was the target too far relative to the move? Was profit taken too early? Did the filter skip good opportunities?

An Agent can get a role, gather data and apply judgment according to your instructions. You can define what matters in the analysis, how to weigh conflicting information, and when to wait or stay out of a trade.

You can start from a ready-made prompt example and adapt it to your approach. Then you run it, review its decisions and improve the instructions.

The level of independence is chosen separately: the Agent can present an analysis, make a decision that waits for approval, or go on to automatic execution within defined limits.

Reactor lets you turn a plan into trading orders. You can set the entry, allocation, protection and profit targets, and also build a staged entry or several linked orders.

After entry, you can give a management Agent a separate task: check the position, keep the protection in place, take partial profit, or consider an exit when conditions change.

You can connect several Agents so the output of one feeds the next: a scanner finds candidates, a decision Agent reviews them, an execution step handles orders, and management and reporting Agents keep tracking.

You can also bring in signals from other systems. The process can run at set times or in response to an event, with the level of involvement you chose.