AI and the Future of Finance: Can ChatGPT Beat the Markets?

Analyzing the potential of artificial intelligence in predicting market movements and gauging market sentiment

The rapid advancements in artificial intelligence (AI) have led to numerous breakthroughs, including the development of ChatGPT and another at least 10 AI models. This has piqued the interest of investors and financial professionals, who are curious if AI can help them make money by outperforming the market. While AI has yet to reach this level of sophistication, recent studies have shown promising results in utilizing AI for market sentiment analysis.

In Subutai, the AI division of Belobaba, we’ve been working on it for the last years with very good results. 

For example, these are the graphs for BTC from last week:

No hay texto alternativo para esta imagen
No hay texto alternativo para esta imagen

Limited Data and Market Unpredictability.

The current limitations of AI in predicting market movements stem from the inherent constraints of pricing and market data. Unlike the vast amount of data used to train large language models, market data is limited in size, making it difficult for AI models to achieve a high level of predictive accuracy. Additionally, markets are inherently unpredictable, driven by emotions, real-time decisions, and ever-changing political climates that are difficult to quantify.

The Growing Potential of AI in Finance:

Despite these limitations, two recent academic papers have demonstrated the potential of Ai models in gauging market sentiment at a level similar to human analysts:

  1. The first paper showed that ChatGPT successfully predicted market movements based on Federal Reserve statements. The AI model’s performance improved even further when trained on historical examples, such as rate hikes by Jerome Powell.

You can find this article here:

The problem here is the time it takes to achieve the conclusions, so you can take advantage of the market. Personally, I think it’s difficult to get real results from this. 

No hay texto alternativo para esta imagen

2. The second paper tasked ChatGPT with interpreting corporate news headlines. The AI model’s interpretations were not only accurate but also correlated with the stock’s subsequent moves, showing potential for investment decision-making.

You can see the article published 9 days ago here:

They found that more basic models such as GPT-1, GPT-2, and BERT (Google) cannot accurately forecast returns. A complex model is needed for this accuracy.

And of course, a strong trading strategy is required to take profit from this kind of analysis.

No hay texto alternativo para esta imagen

The Future of AI in Finance:

While it is unlikely that AI models like ChatGPT will outperform seasoned investors like Warren Buffett in the near future, they are increasingly emerging as valuable tools for both quantitative analysts and retail investors. By parsing through vast amounts of information and providing insights into market sentiment, AI has the potential to revolutionize the financial industry.

As AI continues to evolve, investors and financial professionals can expect more sophisticated tools and models that will enhance their decision-making processes. Although AI may not be able to predict market movements with absolute certainty, its growing capabilities in sentiment analysis and information processing are making it an indispensable tool in the world of finance.

Belobaba launched Subutai in May 2021. We are in the right place at the right time. 

Yours in crypto and Ai.