A recent crypto trading experiment revealed that lesser-known AI models significantly outperformed popular ones. The AI models qn and DeepSeek were the top performers, generating nearly 100% and 80% profit, respectively, over a two-week period. Conversely, the most well-known models, ChatGPT and Gemini, resulted in significant financial losses. This outcome serves as a strong caution against using popular, general-purpose AI for managing investment portfolios without understanding the high risks involved. Investors exploring AI-driven trading should consider researching specialized models rather than assuming mainstream AIs are effective for financial markets.

AI-Powered Crypto Trading

• A real-world trading competition was conducted using six major AI models (ChatGPT, Grok, DeepSeek, Claude, Gemini, and qn) to trade cryptocurrency with real capital.
• The experiment, which ran for almost two weeks, revealed surprising results about which AI models were most effective for trading.
• The most well-known, mainstream AI models were not the most profitable.

Takeaways

• The primary insight is that not all AI models are suitable for financial trading, even if they are powerful in other areas.
• An AI's general popularity or capabilities do not necessarily translate to success in the highly specialized and volatile crypto market.
• The results demonstrate the high-risk nature of using AI for trading, with outcomes ranging from nearly 100% profit to significant losses in a short period.

Top Performing AI Models

• DeepSeek: This model performed exceptionally well, achieving an 80% profit during the competition.
• qn: This model was the top performer, generating nearly a 100% profit.
• Claude & Grok: These models were described as "barely in profits." It was noted that Grok had a strong start but its performance faded over the two weeks.

Takeaways

• For investors exploring AI-driven trading, this experiment suggests that lesser-known or more specialized models like DeepSeek and qn may offer better performance than their mainstream competitors.
• The strong performance of these specific models could be an area of interest for further research, but it is based on a very short-term and limited experiment.

Underperforming AI Models

• ChatGPT & Gemini: These two models, arguably the most famous and widely used, were the worst performers in the trading challenge.
  • Both models ended "deep in the red," indicating they lost a substantial amount of the trading capital.
  • The podcast host highlighted the shocking result that the models generally considered the "best" were the ones that would have lost investors the most money.

Takeaways

• Warning for Investors: This serves as a strong cautionary tale. Do not assume that a popular, general-purpose AI like ChatGPT or Gemini can be used to successfully manage your investment portfolio without significant risk of loss.
• Risk Factor: The failure of these top-tier models underscores the immense risk and complexity of automated trading. Their poor performance highlights that a deep, specialized understanding of market dynamics is crucial, which these general models may lack.