Bitcoin-Color 4200: Candlestick Color Prediction Dataset Using Ichimoku Signals and Additional Features

The Bitcoin-Color 4200 dataset comprises 4,281 daily samples from the Bitcoin 24h chart between January, 2012 to July, 2023, created for a binary classification problem to predict the color of the next day’s candlestick—Red (price decrease) or Green (price increase). Each sample includes 28 input features, primarily derived from Ichimoku Clouds, such as the Conversion Line, Base Line, Leading Span A & B, and Lagging Span (Tenkan-sen, Kijun-sen, Senkou Span A, Senkou Span B, Chikou Span), alongside other suggested features. The output is a binary label indicating the predicted candlestick color, enabling machine learning models to forecast daily price movements. Candlestick color prediction is essential in trading and technical analysis, as it helps traders anticipate price direction, optimize strategies, and manage risk. The Ichimoku system, with its ability to capture trends, momentum, and key support/resistance zones, is central to this dataset, offering powerful insights into market behavior. By combining these signals with additional technical indicators, the dataset provides enhanced predictive capabilities, making it an excellent resource for research in financial forecasting, algorithmic trading, and cryptocurrency market analysis.