Binance AI Pro Turns Trading Instructions Into Simulated Strategies

Binance AI Pro can convert plain-language trading instructions into adjustable workflows, with a demonstrated process that lets users test a strategy in a simulated account before deciding whether to run it live.

The product is part of Binance Intelligence, which also includes Binance AI for general users and AgentOS for developers. AI Pro lets users create trading strategies through natural-language conversations.

The demonstrated workflow follows three stages: prompt, test and trade. Users can describe entry conditions, indicators, position sizing and risk controls, then revise the resulting strategy before simulated execution.

In one test, AI Pro was asked how to allocate 50 Bitcoin (BTC) through Binance Earn while avoiding the risk of losing the principal. AI Pro read the user’s account balance, identified that the account did not contain 50 BTC and excluded products that could change the amount of BTC held. It prioritized flexible and fixed-term BTC products.

The test also asked AI Pro to create a BTCUSDT perpetual contract long strategy. The conditions required the one-hour and four-hour charts to remain above the 20-period moving average, while the 15-minute chart had to reclaim that average with trading value exceeding 1.3 times the average of the previous 20 periods.

The strategy further required open interest to rise more than 5% over one hour, the active buy-sell ratio to exceed 1 and the funding rate to remain below 0.01%. The requested position size was 10%, with 2x leverage, a 1.5% stop-loss and a 3% take-profit.

AI Pro separated the conditions by function, using longer time frames to assess direction and shorter intervals to identify entries. It also sought clarification on whether the 10% position referred to notional exposure or margin, along with settings for isolated margin, one-way mode and checking frequency.

The strategy was then placed in a simulated account so its conditions could continue to be checked against real-market conditions. The test stopped short of live execution.