499 AI in Trading - Can It Really Build Reliable Trading Systems?

Want to learn more about systematic trading? Click here to watch the FREE Masterclass!Artificial Intelligence can write code, analyze vast amounts of information, and generate ideas in seconds. But can it also build a truly reliable Trading System that is ready to go live? Today, tools such as ChatGPT, Claude, and Gemini seem to offer enormous potential for trading. We can ask AI to develop a strategy, write the corresponding code, and even generate backtest results. However, this does not necessarily mean that the results are accurate.At Unger Academy, we have tested Artificial Intelligence for developing Trading Systems and tools designed to support traders in their work. The results were interesting, but they also revealed some significant limitations. In one of our first tests, for example, we asked AI to develop an intraday strategy for Crude Oil.The system returned seemingly accurate data and performance metrics. However, when we analyzed the results, we discovered that the backtest had been conducted using data that did not match the market's actual historical performance.This is precisely the problem: AI can produce convincing answers even when the information it uses is incorrect. In trading, relying on unverified results can be particularly dangerous. A seemingly profitable strategy could be based on inaccurate data, unrealistic trading conditions, or calculations that fail to account for commissions, slippage, and liquidity.This does not mean that AI in trading is useless. On the contrary, it can be an extremely powerful tool for generating ideas, writing and debugging code, automating certain tasks, and accelerating the development of new tools. However, using AI in trading correctly requires the skills needed to formulate the right prompts, verify the results, and identify potential errors. Moreover, a single Trading System, even a valid one, is not a complete trading solution. Building a more robust approach requires diversified strategies, reliable data, testing, validation, risk management, and a structured development process.Artificial Intelligence can therefore support traders and make their work more efficient, but it cannot replace knowledge, methodology, and experience. In this video, Andrea Unger, 4-Time World Trading Champion, explains what AI can currently do in trading, its main limitations, and how it can be used within a controlled and informed development process. Enjoy!