Utilized time-series analysis from three different methods (LSTM, Prophet, and Auto-ARIMA) to predict the closing price of Bitcoin between 2015 and 2017 to the end of 2020.
- ARIMA is a very popular time-series analysis tool which combines AR and MA while correcting for stationarity.
- Prophet is a library released by Facebook which produces reliable forecasts through combining many reliable methods, including ARIMA, as well as introducing new aspects. It is particularly effective in handling outliers, shifts in trend, and missing data.
- LSTM is an artificial recurrent neural network often used in Deep Learning which can also be utilized for time-series analysis
It is designed as a easily readable, comprehensive teaching tool for a lecture series in algorithmic trading for Investment Club. The scripts are written in IPython Notebooks, which make them clear and easily legible while including visualizations generated from Python.
Tools Used: Python, Jupyter Notebook, NumPy, Pandas, Keras, Matplotlib, TensorFlow, Scikit-learn, Statsmodels, Plotly, and PyPlot
 |
 |
 |
 |