Cryptocurrencies trading bot

The goal of this project is to build a parametrized bot to automate trading based on indicators, statistics and machine/deep/reinforce learning.

This is a work in progress project, not production ready.

Setup

I strongly recommend to use a python virtual environment with every project that you are working to avoid problems with python and library versions. See here.

Install Python dependencies:

Run the following line in the terminal: pip install -r requirements.txt.

Create a Binance account and api keys:

(Currently the project only support Binance, but you might create the functions needed to other exchanges.)

• You could use my referral link to create a binance account, if you want.
• Create a new api key. Read here for help.

It is recommended to use the Binance Spot Test Network before to avoid mistakes and lose your funds. You could follow these instructions to do this.

Create a bot in Telegram:

To receive messages or alerts you need a bot in Telegram. Follow these instructions.

Create user configuration:

Create a config file named config.yml based on config.yml.example, then update with your own data. The configuration file consists of the following fields:

• params:

  mode: 1 - interactive, 2 - process, when you want to run as a service or in background;

  symbol: a list of pairs to trade. Example: ['BTCUSDT', 'ETHUSDT']

  interval: an interval to trade. Valid intervals are: 1m, 3m, 5m, 15m, 30m, 1h, 2h, 4h, 6h, 8h, 12h, 1d, 3d, 1w, 1M

  base_asset_order: List of the base assets to create na order. Sometimes you want analyze and compute data from a different symbol. The list must be the same lenght as 'symbol'. Example: ['BTC', 'ETH']

  quote_asset_order: List of quote assets to create orders. The logic is the same as 'base_asset_order'. Example: ['USDT', 'USDT']

  live_trade: live or test trade: 0 - False, 1 - True. Be careful with this parameter.

  roll_oco_orders: recreate oco orders with new values when they are going to come closer to the limit price: 0 - False, 1 - True.

• api_creds:

  binance1_access_code: Binance API key generated in the Binance account setup stage.

  binance1_secret_code: Binance secret key generated in the Binance account setup stage.

• api_test_creds:

  binance_test_access_code: Binance Spot Test Network api key.

  binance_test_secret_code: Binance Spot Test Network secret key.

• telegram:

  bot_token: token from telegram bot

  bot_chatID: chat id

• api_glassnode:

  api_glassnode_key: api-key from glassnode to gather some on-chain indicators. A Standard Plan (free) is sufficient.

Create trade strategies:

Copy trade-strategies.csv.sample to trade-strategies.csv and update it with your strategies. Be careful with CreateOrders column!

• Symbol: symbol to trade and calculate statistics/strategy;
• Interval: strategy interval;
• SignalColumnName: column name in the dataframe that has the signal to buy or sell. See the function update_signal_by_strategy in the trade_utils.py file what strategies can be used.
• CreateOrders: 0 - only send a message, 1 - create orders also.
• IsPercentBuy: 0 - buy amount in value, 1 - buy amount in percentual.
• BuyAmount: value to buy, if percent, value to buy 1.0 = 100%.
• IsPercentSell: 0 - sell amount in value, 1 - sell amount in percentual.
• SellAmount: value to sell, if percent, value to sell 1.0 = 100%.
• OCOStrategy: 0 - don't create an oco order automatically with the oposite side; 1 - create an oco order in addiction to an order created by strategy.
• Message: message that was sent when the signal to sell or buy is reached.

Run

python trade.py

Jupyter notebooks files

There are also some jupyter notebooks to perform analysis, backtesting and predictions:

• cryptocurrencies_analysis: price analysis and backtesting on strategies;
• cryptocurrencies_ml: a basic machine learning model;
• cryptocurrencies_alts: comparative altcoins analysis;
• cryptocurrencies_arbitrage.ipynb: profitable arbitrage scan;
• cryptocurrencies_on-chain_analysis.ipynb: on-chain analysis. You need a glassnode api-key from a Standard Plan (free) to execute the entire notebook.

Don't forget to install the jupyter notebook in your environment before:

pip install notebook