This study examines the use of artificial intelligence and machine learning techniques in cryptocurrency price prediction through a comprehensive literature review. Cryptocurrencies, emerging as a natural consequence of the digital transformation process, are financial assets based entirely on data, and in this data-driven ecosystem, artificial intelligence algorithms have become indispensable tools for predicting future price movements.
The literature review reveals that a consensus has not yet been established in cryptocurrency price prediction modeling. There are conflicting findings regarding the superiority of different machine learning models such as LSTM, XGBoost, ANFIS, and SVR. While some research argues that LSTM is the most reliable tool in highly volatile markets, others claim that ensemble learning methods like XGBoost are more successful in practical trading optimization.
There are also serious debates regarding the data sources to be included in prediction models. Different views exist on the relative importance of social media sentiment, technical indicators, and macroeconomic variables. While some studies show that social media data is strong in short-term predictions, others emphasize the determining role of technical analysis indicators.
Model explainability and security issues have also gained importance in the literature. It is emphasized that explainable artificial intelligence techniques such as SHAP analysis are essential for model transparency, and that machine learning should be used in manipulation and anomaly detection. The study demonstrates that more empirical research and theoretical frameworks are needed in this field.