How AIZEN Works
Input Data: AIZEN’s predictions are based solely on technical market data, such as historical price movements, volume and calculated technical indicators. Current model does not rely on sentiment analysis or external factors like news or social media trends.
Transformer Architecture: Similar to other AI models like ChatGPT, AIZEN uses a transformer model, but instead of processing text, it predicts future price movements by analyzing technical data and identifying patterns and relationships within it.
Temporal Fusion Transformer (TFT): combines the best aspects of autoregressive models, RNNs and attention models that allow the model to process sequences of market data over time, learning how past movements and indicators relate to future outcomes
Prediction Output: AIZEN is designed to predict price movements over 72-hour timeframes. It focuses on short-term predictions, but it has the flexibility to generate insights across different time horizons, with plans to expand this capability in the future.
Breadth of Coverage: AIZEN currently monitors and predicts price movements for 64 preselected cryptocurrency tokens. As development continues, the system will expand to cover additional assets.
Last updated
