RIALTO.AI Trade Bot Infrastructure

At RIALTO.AI, we are developing a trade bot for automated prediction trading. Prediction trading is all about determining the right time to either buy or sell a cryptocurrency. We have implemented and backtested several strategies so far, and we continue working on the development of the new ones. Our backtesting process is extremely rigorous, and only the most profitable strategies make it into the production environment.

THE INFRASTRUCTURE

The trade bot infrastructure consists of four parts: real-time stream processing engine, historical data service, backtesting environment, and the trade bot.

The real-time stream processing engine is responsible for running the trading strategies. The input for strategies is real-time data obtained from the exchanges and the historical data from our databases. Strategies analyze these inputs and generate signals which are interpreted by the trade bot.

The historical data service is storing trade-related data to the databases. This data is then used for backtesting where it is used to evaluate the profitability of the strategies, and in the real-time stream processing engine where it is fed to strategies which require historical data in addition to the real-time data.

The backtesting environment was developed from scratch since we could not find a solution that would meet our needs. Usually, such environments support only “single-strategy” testing, while our goal was to improve this and have the option to perform testing in a “multi-strategy” backtesting environment. We started with the assumption that by interleaving the strategies we could achieve greater profitability and thanks to our multi-strategy backtesting environment we were able to confirm this.

The last part of our infrastructure is the trade bot which performs the trades on the exchanges according to the signals it receives from the strategies.

THE DEVELOPMENT

The process of designing and developing such a complex and advanced infrastructure has been lengthy and time-consuming. However, we have achieved huge improvements in the reliability (in terms of uptime) and scalability (new strategies can be tested and implemented with little effort). Additionally, tradebots are built in a modular way so that they can integrate with the RIALTO.AI platform that enables users to replicate all trades for their own account with a single click. In contrast to many other social trading websites, we are focusing on quality over quantity thus we will deploy only best performing algorithms. We are making a significant step forward with the multi-strategy trading by deploying a “long/short” strategy, which performs well in both, bullish and bearish markets.

Twitter | Medium | Facebook | Linkedin | Telegram