Forex-Trading

A simple implementation of HFT (High-Frequency Trading) in Python on the concept of DQN

Goal :

• Finding patterns for avoiding spread loss and limiting the order book
• Automating the trading by certain default values (you can find the config file inside the file) by using the DQN model.
• Making a model that can handle multiple orders without avoiding conflict with the analysis
• Avoid using any extra library or API

Some basic information:

• Support tensorboard
• Written in TensorFlow v1
• if you using TensorFlow v2 then import this on top of the code to avoid conflict of depreciated function like random_seed_set and session

import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()

• HFT (High-Frequency Trading): High-frequency trading (HFT) is a type of algorithmic financial trading characterized by high speeds, high turnover rates, and high order-to-trade ratios that leverages high-frequency financial data and electronic trading tools.
• More info: https://www.investopedia.com/terms/h/high-frequency-trading.asp

Training Data

• 1000 Episodes
• EURUSD (5 min interval - 2 Year Period)
• GBPUSD (5 min interval -2 Year Period)

Install

• Check config file based on your requirements
• Install lib

pip install -r requirements.txt

Training

python3 index.py

• Code Documentation

python3 index.py -h

• Tensorborad results link

https://tensorboard.dev/experiment/H86sR9cmRlOUvRV8yBwLEA/

Results

• Map
• Result

Remarks

• Results will be different if you are using leverage (for testing purpose 50x leverage is used) and charge based on your broker
• IT'S A EXPERIMENT [NO RESPONSIBILITY FOR ANY INTEGRATION IN THE LIVE MARKET]