The author discusses their pursuit of using AI for accurate trading signals, highlighting challenges such as inconsistency, limited context windows, difficult backtesting, and high costs. Asking AI models like ChatGPT for specific trading advice proves ineffective due to their lack of structured strategy and historical data limitations. To address these issues, the author proposes a hybrid approach where AI acts as a conductor, running simulations on powerful AMD EPYC servers. This method allows the system to backtest, identify optimal parameters, and adapt to market changes after each trade. The result is an integrated system that utilizes LLMs for strategic reasoning and CPU power for processing. Notably, the system, called TrendFi, is designed for trend-based modeling rather than day trading, focusing on identifying significant trend shifts and providing alerts for those moments. For more information, visit TrendFi at https://trend.fi.
Show HN: Introducing TrendFi – AI-Powered Trading Signals with Self-Optimization
