A Data Driven Cryptocurrency Trading Bot

You will probably not make money with this, you are at your own risk if you choose to put in real money!
This program is built with simplicity in mind, you can train/test/run a model for a certain asset in less than 30 minutes.
This program runs on Coinbase Pro, if you would like to use it as an investment tool, be sure to have a Coinbase Pro account and insert all necessary API information in authCredentials.py.
This program uses 4 different machine learning algorithms to generate buy and sell signals for a given asset and time interval.
A pre-trained model for BTC/USD is available and ready to be used.

How it works

The file subroutines.py contains all the methods used to calculate the different technical indicators of a given asset.
The file gen_transform.py reduces the dimensionality of the dataset by selecting the top 200 features using Random Forest.
The reduced dataset is then used to train 4 different machine learning algorithms, K-nearest neighbor classifier, random forest classifier, gaussian naive bayes classifier and gradient boosting classifier.
The 4 models' outputs are combined using a weighted average, and the final outputs are used as raw predictions. Below shows the raw predictions on the BTC/USD pair.

Historic prices are first transformed into chunks of equal sizes, the minimum and maximum for each chunk is considered a buy and sell label respectively.
To visualize the profit and percent gains for a large range of chunk sizes, execute the line python general_test.py
Different assets often require different chunk sizes, the default chunk size is 320.

The raw outputs from the ensembled model have too many buy/sell signals in the same general area.
To combat this, every time a buy signal is received, it won't immediately trigger a buy action, but rather sets up a stop-loss and take-profit margin that centers at the previous closing price.
The margins are set up according to the risk tolerance and multiplier settings in trader.py.
If a new buy signal is received before price breaks the margin, then a new margin will be set at the previous closing price.
The buy action will only be executed when prices eventually crosses either the stop-loss or take-profit.
The same operation is done on sell signals. Blow shows the same predictions after the prediction processing.

The file trader.py contains the real-time prediction processing methods.
You might see different results when you change the risk tolerance and risk multiplier values.

This program requires the packages sklearn, termcolor, imblearn as well as cbpro.
To install, clone this repo via git clone https://github.com/SnowCheetos/Emsemble-Tradebot.git.
A pre-trained model for BTC/USD is ready to be used. To use the model, execute the line python trader.py and type in BTC for the asset.
Make sure to have inserted all the API information in authCredentials.py for Coinbase Pro.
If you would like to train the model on an asset other than BTC, execute python train.py and enter the asset. Make sure the asset is available for trading on Coinbase Pro (XRP is not).
You are encouraged to change the function parameters to what works best.
You are also encouraged to set up a general stop-loss when you are holding a certain asset. Cryptocurrency prices are extremely volatile and the model can make bad decisions during extreme volatility.

To visualize performances of each model, execute python general_test.py.
To visualize the performance of the ensembled model, execute python ensembled_test.py.