This page updates every 2 minutes, displaying predictions from advanced machine learning models. These models analyze OHCL (open, high, close, low) data, trading volume, sentiment analysis, and data from leading cryptocurrencies to forecast price movements with approximately 70% accuracy.
The goal is to provide real-time access to results from recent state-of-the-art research (see [1] and [2] in the Methodology section below). Current predictions are categorized as "going up," "going down," or "can't say" (no clear pattern detected).
The charts above predict price movements over the next 7 days using daily OHCL data, Twitter sentiment analysis, and data from other top cryptocurrencies. Colored dots indicate expected price movement: Red (down), Green (up), or Grey (uncertain). The blue line shows the predicted percentage change relative to the current price. Predictions utilize deep learning (autoencoders combined with ensembles of LSTMs and MLPs) and feature selection methods. Historical accuracy from 2016 to the present is approximately 70%.
These charts forecast price movements for the next 6 hours based on hourly OHCL data and information from major cryptocurrencies. Predictions follow the same color coding (Red: down, Green: up, Grey: uncertain). Accuracy since 2019 is approximately 67%.
Indicators for 12, 6, 3, and 1.5-hour intervals, generated by combining OHCL data with Natural Language Processing strategies for Twitter sentiment analysis.
Indicators covering intervals of 8, 4, 2, 1, and 0.5 days, utilizing daily OHCL data and Natural Language Processing strategies for Twitter sentiment analysis.
Last tweets about Bitcoin:
Predictors employ hourly machine learning and daily deep learning (autoencoders), following methodologies from these papers: