Member-only story
|ARTIFICIAL INTELLIGENCE|LLMs|BUSINESS|STOCKS|
GPT-InvestAR: LLMs for better investment
From Text to Trade: Could an LLM exploit annual reports to predict stock to buy?
LLMs have been used for so many applications, but could they be used to be able to invest in the stock market? Could an LLM be used to predict stock performance and guide our investment strategy?
AI and stocks: a long relationship
Predicting stocks using machine learning is one of the most suggestive and appealing ideas there can be. This idea gave rise to the concept of algorithm trading in which automated algorithms are used to predict the price of a stock.
Considering the potential payoff one can imagine that so many researchers and companies have dedicated themselves to the idea of producing models that can predict the value of a stock. So everything from convolutional networks to LSTM has been tried. In general, it can be said that the classic algorithm design approach is to analyze measurable quantities (stock price, historical trades) to be able to capture patterns and then predict these patterns in the future.
To reduce the difficulty of the tasks, many of these approaches have focused on short-term patterns that are, however, influenced by news. With the advent of Long Short-Term Memory (LSTM), attempts have also been made to develop patterns with more temporal distance.
After all, despite their volatility, these are not randomly generated numbers. On the other hand, the fact that they are affected by news, investor sentiment, potential black swans, and other factors makes them difficult to predict. That is why many investors focus on analyzing not only tangible assets, and financial statements, but also consumer behavior, news, and a whole range of other information.
But this information is textual and needs to be analyzed by models that can not only process this information but also succeed in analyzing it. With the explosion of large language models, it was immediately…
