Artificial intelligence is being hailed as the next big thing in both analytics and art. However, when it comes to crypto trading, the average “crypto knowledge” can outperform AI.
Not a market analyst
The latest data from the CoinGlass AI Model Trading Leaderboard shows how chaotic situations can occur when general-purpose AI systems are introduced into volatile market environments. According to the leaderboard, GPT-5, one of the most advanced language models, posted an incredible return of -64.22%, essentially wiping out most of its trading portfolio and finishing in last place. It experienced the fastest and steepest collapse of all participating models, resulting in a total loss of $6,367.
While other companies such as Grok and DeepSeek have been able to report modest returns (+3 and 64%, respectively), GPT-5's performance clearly shows that trading skill is not just a function of intelligence. Large-scale language models (LLMs) are designed primarily for text generation, reasoning, and problem solving, not for analyzing high-frequency strategies or interpreting volatile market data.
Cryptography is too complex for AI
This is the simplest explanation. Active trading environments require a level of live market awareness, reaction timing, and sophisticated risk management, which cannot be achieved in these environments because the training data is static and historical. Specialized AI systems used by hedge funds and companies like BlackRock, on the other hand, rely on continuous retraining using real-time data and carefully selected domain-specific statistics. These models can identify macro trends, arbitrage windows, and order book imbalances that are not possible with general-purpose LLMs.
Leaderboards reflect the near-random nature of AI trading results, which is like letting a monkey manage your fund. Overfitting patterns can quickly send some models into losses, while short-term volatility can help some models make profitable trades.
If this leaderboard is any indication, you should not trust GPT-5 or its peers with your cryptocurrency wallet. LLM is a great tool for reasoning, text generation, and learning, but when it comes to market execution, you're more of a gambler than a strategist. They were gamblers who lost everything before anyone else in the GPT-5 incident.
