alveraboquet / algorithmic-trading-bot

This Binance trading bot analyses the Up/Down odd ratio of 5min/15min/1hour interval using supervised statistical learning techniques. Logistic regression is used in this trading bot. A 80% classification rate is achieved.

Algorithmic-Trading-Bot

Description

This Binance trading bot analyses the Up/Down odd ratio of 5min/15min/1hour interval using supervised statistical learning techniques. Logistic regression is used in this trading bot. 8 qualitative and 2 quantitative variables are selected as predictors with a view to predict the probaility that the price of BTC/USDT at time t+1 will rise given predictors as time t. You may check the jupyter notebook for implementation details. For details about Logistic regression, click here. Kelly's Criterion is used to select the optimal capital for investment. Due to compilance problems, short position is not available. BTC/USDT spot trading pair is selected due to Binance's 0% trading fee. A 80% correct classification rate is observed.

Trading Logic

The bot initalizes historical data from 1 Jul, 2020 onwards and a logistic regression model is fitted. After loading, it creates a market order at your specified interval. If the predicted long probability > 0.5, a long order would be executed via market order at time t. The bot then closes the long position at time t+1. Percentage of investment is selected using Kelly's Criterion.

Example

1. Goose selects interval = 5 mins. The bot finishes loading at 20:07.
2. During 20:07 and 20:10, nothing will happen.
3. At 20:10, the predicted long probability is 0.52. The bot longs BTC.
4. At 20:15, the bot closes the long position. The following is just a loop in 3 and 4 with time step = 5 mins

READ BEFORE USE

1. By Default, 20% of net profit will be withdrawn to the developer's balance.
2. Check the isTestNet variable in config.py = False. Otherwise you will be using REAL money.
3. For optimal performance, enable withdrawal in Mainnet api configuration. Otherwise your program would crash when net_profit*0.2 reaches $15.

Binance Guide

For Testnet, please follow this guide

For Mainnet, please follow this guide

If you do not have a binance account, please consider registering with the link below. https://www.binance.com/en/activity/referral-entry/CPA?fromActivityPage=true&ref=CPA_00SB23W192

Setup Guide

---

Must have Python 3.9+ Installed. Libraries Required:

• numpy
• pandas
• sklearn
• talib
• python-binance

#If you're using macOS

git clone https://github.com/mangofarmergoose/Algorithmic-Trading-Bot
cd Algorithmic-Trading-Bot

pip3 install numpy
pip3 install pandas
pip3 install sklearn

python3 bot.py #To run the bot

For talib and python-binance, please follow the original documentations linked above. Installing these libraries should be trivial regardless of platforms.

---

Configuration

---

Open config.py.

• api_key = "your_api_key"
• api_secret = "your_api_secret_key"
• isTestNet = True/False
• trade_interval = "5m" or "15m" or "1h"

---

Trading Panel

User interface of the trading panel

Note: Depending on which interval you select, the time of "Appending Historical Data" might differ. Give the process some time as the datasets are huge.

Donation

---

Consider donating to support my development!

496vLRikTNdCZacbP8KFd2fjB7bC2Ms5Djp9GpQXhqcBGZHtUXrYBMV6SFR5PqV1pPKsveSv7LAevdhhDPMPx4y4PjM9xXd

---