Bitget CEO Gracy Chen: The Advantages and Limitations of AI Trading Robots
Original: Bitget exec discusses limitations and benefits of CTA AI trading bot, Original Author: ZHIYUAN SUN.
Recently, Gracy Chen, the General Manager of the cryptocurrency exchange Bitget, detailed the advantages and limitations of the Commodity Trading Advisor (CTA) AI strategy in an interview with Cointelegraph.
One area where AI technology has rapidly integrated with blockchain is cryptocurrency exchanges. Since the beginning of the year, Binance has launched an AI-driven generator for authenticated users to create non-fungible tokens (NFTs). Meanwhile, OKX has introduced AI integration for monitoring market volatility. Bybit has also integrated AI-driven trading tools using ChatGPT.
Starting from June, Bitget has launched a series of AI trading robots. On July 27th, the exchange introduced a new CTA AI robot as part of its AI trading plans for this year, making it the third one released.
Gracy Chen, the General Manager of Bitget, explained the benefits and risks in an interview with Cointelegraph.
Cointelegraph: What is the difference between CTA AI strategy and regular commodity trading algorithms?
Gracy Chen (GC): The AI robot incorporates MACD (Moving Average Convergence Divergence) and Bollinger Bands strategies. It continuously receives historical strategy data, analyzes and processes it, and implements self-learning to output new logical strategies. Therefore, with just simple return rate figures and price charts, AI strategy eliminates the need for complex parameter inputs in algorithms, helping users to select and create strategies more intuitively.
CT: Despite performing well under normal circumstances, AI models often exhibit instability during sudden events such as rapid price increases or decreases. Are there any safeguard measures applicable to users in this regard?
GC: This is undoubtedly a huge impact that any trading platform will face. In particular, the greatest impact on the revenue of CTA AI strategies is that we are prone to receiving many false trading signals during operation, leading to potential losses. However, we protect the interests of users in two ways: The AI strategies we introduce are based on large candlestick chart periods (with a minimum period of 1 hour). Therefore, many abnormal fluctuations in shorter time periods are smoothed out in larger time periods, effectively reducing the impact of false signals. Additionally, we also provide users with options for taking profit and stop loss in the advanced settings, which can automatically help users secure their profits and limit losses to protect their account equity.
CT: Considering that CTA strategies are mainly used for trading commodities on exchanges, such as soybeans or crude oil, how is this strategy particularly applicable to cryptocurrencies?
GC: In principle, CTA strategies grasp market fluctuations based on the relationship between trading volume and price. They are more effective in more volatile markets, such as the cryptocurrency market. Due to advanced technology, progress of the era, and the diversity of participants from around the world, cryptocurrencies are more prone to significant fluctuations.
CT: In the previous discussion, you mentioned that multiple departments at Bitget are attempting to use AI. Can you give a specific example?
GC: We use AI technology to train, analyze, and process samples based on different users' trading habits, and provide intelligent recommendations for different user groups. Additionally, we also use AI to perform some manual tasks, such as generating posters, writing copy, and writing simple code.
CT: What are the advantages of AI-based trading compared to human or algorithmic trading methods?
GC: AI strategies can help users intuitively choose and create trading strategies through simple return rate numbers and price charts, eliminating the need to fill in complex parameters (such as algorithms).
Chen also explained that Bitget will learn from the success of large language models (LLMs) such as ChatGPT to improve its AI robots. "We know that the success of ChatGPT is mainly attributed to two aspects: large-scale sample data and intelligent learning models. Our AI strategies will also start from these two aspects to improve profitability," she said.
