AI Trading Platform

Published IEEE paper: https://ieeexplore.ieee.org/document/9734127

What is the project?

To build a platform for creating, backtesting and paper trading new automated strategies on real-time, minute scale data with the help of AI models.

Components in the project

Data Feeder

• Used to source real-time minute scale data from over 5000 stocks in the stock market.

• Perform very fast, basic calculations on the raw data.

Database

• We use a database to store the data from the data feeder.

• We also use it to perform more complex calculations (such as indicators) very quickly at a large scale for all stocks whose data has been sent from the data feeder.

Strategies

This is the heart of the project

• Strategies read data from the database and can use multiple indicators, machine learning models, or a combination to generate orders according to the rules mentioned in the strategy.

• There may be multiple strategies created and deployed in this service.

Backtester

• This service helps in building Strategies. It allows us test how a strategy would have performed over past data over previous weeks, months and years.

• It takes a parameterised strategy and data over the period of the back test as input. It generates orders over the previous data according to the strategy's rules and outputs a detailed report on the performance of the strategy.

Paper Trader

• This is used to test strategies in real-time market scenarios after a strategy has performed well in back testing.

• It tracks and evaluates orders generated by multiple strategies in real-time market conditions and provides various performance metrics for both strategies and orders.

UI

• This service is a website which provides a GUI for the user to interact with all the other services.

• It allows the user flexibility to create, back test, paper trade and deploy new strategies on real-time data from the stock market.

To build this project

Development environment

1. run these commands to initialize the project:

git clone "git@github.com:Varun487/CapstoneProject_AITradingPlatform.git"

cd CapstoneProject_AITradingPlatform

docker-compose -f devops/docker-compose.dev.yml up

2. To start the backend and frontend development servers + run all containers:

docker-compose -f devops/docker-compose.dev.yml up

3. To stop and destroy all containers:

docker-compose -f devops/docker-compose.dev.yml down

4. Run with fresh build:

docker-compose -f devops/docker-compose.dev.yml down

docker-compose -f devops/docker-compose.dev.yml up --build

5. To create a superuser for the database

docker-compose -f devops/docker-compose.dev.yml run restapi python3 manage.py createsuperuser

6. To run all restapi tests

docker-compose -f devops/docker-compose.dev.yml run restapi python3 manage.py test

7. Command to Initialize the database

docker-compose -f devops/docker-compose.dev.yml run restapi python3 manage.py initialize

8. Command to run all paper trade services

docker-compose -f devops/docker-compose.dev.yml run restapi python3 manage.py papertrade

9. Command to extract and store all data from DB in a csv format

docker-compose -f devops/docker-compose.dev.yml run restapi python3 manage.py extractDBdata

NOTE: To run in production, the commands are the same, but the file is docker-compose.prod.yml

Project members

All members are from Semester 6 Pes University EC Campus

1. Varun Seshu - PES2201800074
2. Hritik Shanbhag - PES2201800082
3. Disha Venkatesh - PES2201800109
4. Samrudhi R Rao - PES2201800126

Future prospects

We intend to build a trading system with a broker to place orders generated by the strategies in the live market and generate consistent profits.