大模型投资竞赛，中国AI包揽前二，GPT-5亏损超62%垫底

Core Insights - The AI model investment competition "Alpha Arena" concluded with two Chinese models, Qwen3 Max and DeepSeek Chat V3.1, taking the top two spots, both generating profits, while four leading American models incurred losses, with GPT-5 suffering the most at over 62% loss [2][4][5]. Performance Summary - Qwen3 Max achieved a return of 22.32%, ending with an account balance of $12,232, while DeepSeek Chat V3.1 followed with a return of 4.89% and a balance of $10,489 [3][4]. - The remaining models, including Claude Sonnet 4.5, Grok 4, Gemini 2.5 Pro, and GPT-5, all reported losses exceeding 30%, with GPT-5's balance dropping to $3,734 [5][6]. Trading Strategies - DeepSeek's stable performance is attributed to its "professional alignment" as it is backed by a quantitative firm, employing a straightforward strategy without frequent trading or stop-loss measures [8]. - Qwen3 Max utilized an aggressive "All in" strategy on a single asset with high leverage, which, despite previous losses, resulted in the highest profit [8]. - Grok 4 exhibited a high-frequency trading style with significant volatility, while Claude Sonnet was noted for its analytical prowess but suffered from indecision in trading [8][9]. Market Dynamics - The competition highlighted the unpredictable nature of real market trading, where even advanced AI models struggle to maintain consistent returns [6]. - The event was initiated by Nof1 to explore the potential of AI in financial markets, suggesting that the financial sector could serve as a challenging training ground for AI development [9][10].