Chinese AI Models Gain an Advantage in Crypto Trading

Chinese artificial intelligence models DeepSeek and Qwen AI outperformed their American competitors in a cryptocurrency trading competition involving real money. This development was confirmed by the AI research laboratory focused on financial markets, Nof1.

Nof1 launched a competition called Alpha Arena on October 17, which tests the investment capabilities of popular AI models. The models were directed to trade cryptocurrencies through the decentralized exchange Hyperliquid with an initial capital of $10,000, with questions and datasets provided in the same way initially.

During the competition, DeepSeek V3.1 Chat increased its initial capital of $10,000 to $21,600, achieving a gain of %116. The Qwen 3 Max, developed by Alibaba, also raised its capital by nearly %70, reaching the level of $17,000.

Compared to previously tested models, it emerged that low-performing models like Google's Gemini 2.5 Pro and OpenAI's ChatGPT 5 experienced losses of more than %60. Nof1 noted that these models generally preferred smaller positions, stating that they were less aggressive compared to their competitors.

Some experts suggested that the results of the Alpha Arena competition followed a random walk. This mathematical model describes a path made up of sequential random steps and argues that the average trade position converges toward the starting point.

The Alpha Arena competition is set to conclude on November 3, indicating that significant changes in the rankings may be possible.

Alpha Arena is just one of many trials testing the trading abilities of artificial intelligence models. Last June, researchers from Stanford University reported that a model trained only on public information outperformed %93 of managers over a 30-year period with an average difference of %600.

In July, a user initiated a six-month experiment using ChatGPT to pick small stocks with an initial capital of $100. Last week, the user's capital had declined to $76. During the same period, investments in the S&P 500 would have risen to $109.46.