Automated Trading Model Platform
Future Software Technologies
Semester programme:Complex Software Systems
Client company:Schriek,Erik
Project group members:Roman
Naadir
Owen
Eva
Camiel
Project description
The goal of the project is to create an automated trading platform that independently tests, evaluates and optimizes trading strategies using historic market data to find out whether a strategy is profitable under real market conditions.
Context
Because of the time-consuming nature of manual trading, trading platforms offer to create trading strategies which executes trades based on a set of rules placed by the user on the real market.
To be able to know if your strategy will work u will have to test the strategy on past data but most brokers are not interested in you becoming extremely profitable with your strategy. Their way of making money is often through fees which they generate from each individual trade.
In order to help out the small individual traders we automate the testing of created strategies independent of brokers, so u can reliably know if the strategy is profitable in the real world. Both in past data and in future market conditions that are constantly changing.
To achieve this, we will be using accurate historic data to test our strategies and eventually use all the outputs as a prediction on which strategies would be profitable. For all testing of the strategies we use Meta Trader 4, this is a windows program that allows you to make trades and preform tests on your strategy using historical data.
Results
The project delivers a automated trading strategy testing platform working independently of a broker.
The most important outcomes are:
- The automated historical data pipeline to eliminate manual data preparation. this enables scalable back tests and this reduces manual updating.
- The scalable Meta trader 4 back test system, this system needs to make testing strategies on Meta trader 4 faster, and easier to scale adding more compute power for testing should be as easy as possible.
- the prediction module, this system should test your strategy with multiple parameters and cluster winning parameters together to give an recommendation for improving your testing strategy.
About the project group
Our group consists of 5 Software Engineers.