AI BriefingPredicting Cryptocurrency Prices with ChatGPT
Key point
The study explored the potential for price prediction in the highly volatile cryptocurrency market through ChatGPT's news sentiment analysis.
Details
The cryptocurrency market lacks a clear valuation method, and prices are characterized by being determined by factors external to the market, such as technological developments, regulation, and investor sentiment.
This study used a LLM (Large Language Model) to analyze whether sentiment analysis results from news articles could contribute to cryptocurrency price prediction. To this end, global data from CoinMarketCap and South Korean market data from Upbit were collected.
The news data utilized articles from Naver News and 135 outlets across 22 countries, and to prevent data bias, a preprocessing process using the sBERT model was performed to remove articles with high semantic similarity.
Finally, the study provided ChatGPT with a specific guideline, a prompt, to extract sentiment scores from news articles and used these as variables in a price prediction model.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Stock Price Prediction Using ChatGPTAI That Reads Your Day, Development Story of the mini Diary Service