Automated trading system

About the client

A high-potential fintech startup.

How the project started

The project began at the end of 2024, when the client approached XPG Factor with a clear request: to develop an automated system for trading. Until then, his workflow involved manually monitoring TradingView signals and maintaining records in spreadsheets. So, the process was inefficient and time-intensive.

Our goal was to automate all routine tasks: the system would monitor real-time market signals, open and close trades with multiple brokers and require minimal human involvement.

Key features of the system

The project took six months and resulted in a fully automated trading system that freed the client from manual analysis.

• 24/7 automated market monitoring
  In an environment where signals are unpredictable, any delay is fraught with the risk of missing a trade and destroying the entire strategy. We solved this by implementing robust data synchronization, enabling the system to detect the right moment and determine what trade to make, through which broker, and how.
• Trade distribution
  We built a fully automated mode that monitors signals 24/7, executes trades as soon as predefined conditions are met and adapts to market realities like commission rates and asset availability. If one broker is out of stock, the system simply moves the trade to another.
• Two modes of operation
  We have also added the option to work in two modes: paper trading – for safe testing, and live trading – with real money. This allows the client to first try out strategies without risk, and then switch to the “live” mode.
• Flexible trading strategies
  One of the main challenges was creating strategies with flexible conditions, such as “trade until a +5% PnL is achieved or until 3:30 PM New York time.” There were dozens of similar combinations. So, we built a system that allows the client to easily configure strategy parameters and pause, restart, or modify them whenever needed.
• Full transparency & reports
  One of the key tasks was to ensure complete transparency. That’s why we integrated PnL reports – both for individual trades and aggregated by day, asset, and user. Now the client knows exactly which strategies performed well, where losses happened, and what changes to make.
• Multiuser support
  The final touch of the project was multiuser support, which allowed to expand the system’s capabilities and make it more flexible. Now several trusted users can simultaneously work with one platform, exchange strategies and jointly control trading processes.

Security and privacy

Security was also one of the key aspects of the project. So, the system encrypts all stored data in the database, so even if someone gains unauthorized access to the virtual machines, decrypting the data is impossible without specific keys. These keys are stored exclusively in GitLab’s secret variables and are not accessible on any of the server machines.

At the same time, the system itself cannot withdraw or transfer funds – it only automates order placement with brokers. All financial transactions stay under the full control of the client and the brokerage platform.

Issues we had to solve

During the project, we faced several challenges that required close attention and prompt decisions. First, various brokers responded differently to trading signals, especially in the pre-market hours. Second, limited access to broker accounts and TradingView data meant debugging had to be done in close contact with the client through regular calls and joint analysis.

By optimizing computing resources, we built a flexible architecture that adapts to changing market conditions and ensures stable operation in all scenarios. At the same time, given the client’s desire to stay within GitLab’s free limits, we initially used self-hosted GitLab runners on Google Cloud virtual machines.

Future plans and goals

The next step in our work on this project is to test the upcoming iteration, which is expected to take about six weeks. If the tests go well, the client is thinking about moving the project to a full-time format.

In addition, the project may undergo a comprehensive architectural modernization, including the development of a frontend part for more convenient visualization of deals.