Accuracy of artificial intelligence in predicting individual stock prices based on historical data

As artificial intelligence continues to become more prevalent in society, more people begin to rely on it on a day-to-day basis, sometimes also for financial decisions. Considering money plays a big role in many people’s lives, the accuracy of artificial intelligence models that play the role of being financial advisories becomes critical in safeguarding people’s money. If human financial advisors require a license to legally give financial advice, shouldn’t machines go through equally or more rigorous testing?

This paper investigates a simple AI structure, consisting of six models, and how they are affected by certain variables in dealing with randomness, how well they perform when tested against real-world data and a simulation against real market conditions, which includes one simulation indicating a return of 1115.35% on investment from an initial investment of $5000. It also investigates the hypothesis that well-trained AI models can significantly outperform human stock traders when simulated against market conditions.

This paper found that artificial intelligence is more than capable of predicting the stock market based solely on numerical factors, taking advantage of predictable movements in market structure to make a profit for its investors. This would transform the trading market as we know it as individuals, researchers, and stock-broking firms begin to rely on AI to put in personal investments into the stock market.