LSTM Bitcoin Prediction
2024A deep learning pipeline using LSTM networks to forecast Bitcoin price trends from time-series data. Built in Python with Jupyter.
PythonLSTMMLPyTorch
Overview
Deep learning pipeline built in PyTorch/TensorFlow to analyze financial time-series data and forecast Bitcoin market trends.
The Challenge / Problem
Financial market data contains high volatility and noise, making linear models and traditional regressions inefficient for medium-term forecasting.
The Solution
Created a recurrent neural network (LSTM) using sliding time windows for feature engineering, normalization, and temporal validation.
Project Highlights
- Feature engineering with technical indicators (RSI, MACD, Bollinger Bands)
- LSTM model training with early stopping to prevent overfitting
- Interactive notebooks with real vs predicted data visualizations