Autonomous Trading Systems Inc (ATS) is a Miami-based company focused on productizing high frequency trading (HFT) robots as a service (SaaS). Formed in 2015, its team is comprised of top executives from varying backgrounds (Sony, BTP Global) who have identified a unique opportunity in the electronic cryptocurrency space.
After spending close to 2 years analyzing trends within the various Bitcoin exchanges, ATS had developed a trading strategy that sought to capitalize on certain recurring market inefficiencies. Critical to their strategy was the speed of opportunity discovery and execution, which often occurred in timeframes measured in milliseconds. Given the limitations of a manual approach, ATS required the development of a sophisticated backend in order to automate the discovery, analysis & execution of any given trade. Beyond speed & strategic viability, the system needed to also leverage Machine Learning/AI in order to maintain a competitive edge & ensure continual discovery of new opportunity types over time.
Given that the system was expected to correctly handle 6 figure portfolios and higher, our margin for error was substantially less than with other MVP’s that we had worked on in the past. Traditionally we encourage the release of products that facilitate the build, measure, learn feedback loop, often prioritizing market feedback over “internally perceived perfection”. However with Bitbot, market validation was derived entirely by whether the platform could successfully make positive financial gains using the algorithmic trading strategy that had been agreed upon. Following an extensive requirements gathering phase, the Codelitt team set out to build the first iteration of the platform within 4 weeks. The goal? Generate just one dollar of profit when the system went live.
Generate just one dollar of profit
when the system went live.
High frequency trading is used by investment banks, hedge funds and institutional investors to execute millions of orders and scan multiple markets/ exchanges in a matter of seconds- thus giving the institutions that use the platforms a huge advantage in the open market.
To build a fast and reliable trading bot, we leveraged cutting edge technologies that provided a reliable testing environment, fast execution time, robust error handling and allowed us to introduce changes at the velocity required by this kind of project. Bitbot has a few Ruby/Java micro-services responsible for the trading system and a Ruby on Rails backend application to provide information to the React.js frontend application.
Bitbot is logically designed to work with several components. It uses microservices to follow a layered approach, resulting in different isolated components, with single responsibilities, that can communicate with each other through the use of a Message Broker pattern. The usage of the microservices architecture allows the different components to be isolated and self-contained, which facilitates independent scaling. It also makes possible the choice of the best technologies for each component, since they don’t need to be all written within the same architecture and or language.
The use of the Message Broker pattern makes the robot work in an event-driven system, making it more reactive and predictable. The robot execution is composed of cycles and each cycle is composed by different phases that trigger specific events that are used to orchestrate the flow.
The stack of Bitbot is mainly built using:
Ruby on Rails
As a backend centric platform, the client requested a simple dashboard to monitor the “status/health” of the bot and get a snapshot of its most recent activity. Our development team built out a series of APIs that feed real-time data to a secured web-based frontend which contains data related to:
After the completion of the first development phase in May of 2016, Bitbot executed a succession of micro-transactions, yielding $5 in profit after being live for only 2 minutes. This proof of concept gave ATS the confidence they needed to confirm the viability of the initial trading strategy and move forward with subsequent development iterations- shoring up the platform ahead of an invite to private investors.
Bitbot executed a succession of micro-transactions,
yielding $5 in profit after being live for only 2 minutes