Our customer is an Irish proprietary trading firm. The primary focus of the company lies in trading highly correlated products while capturing minor price discrepancies.
Detailed information about the client cannot be disclosed under the provisions of the NDA.
The client approached Innowise Group with a specific requirement to develop a custom web-based quantitative trading platform. Specifically, they sought a data-driven platform capable of executing their trading strategies for the cryptocurrency market based on a vast amount of historical and current data. They emphasized the need for a system that could incorporate various data sources, such as transaction volumes and alternative data requests.
The client’s previous trading system was not sufficiently responsive to rapidly changing data, so it wasn’t efficient for their needs. It suffered from significant delays, taking 2-3 seconds to process information, which proved to be prohibitively slow for making timely trading decisions.
To successfully implement new trading strategies, a fundamental requirement was a high-speed system capable of processing a substantial volume of financial quotes and other relevant data in real time. Swift identification and analysis of short-term discrepancies between correlated assets were essential since they could emerge and disappear within a matter of seconds. Therefore, the new system had to rapidly capture and process this information within milliseconds to facilitate accurate calculations and execute successful trades.
To address these challenges, our trading software development company embarked on developing a new quantitative trading platform from scratch to meet the demands of a fast, reliable, and custom-built solution.
By analyzing trading volumes and employing ML boosting algorithms, we detected anomalies within the market that indicated favorable buying opportunities. The system leverages Grafana as a powerful tool to query, visualize, alert on, and gain insights into various trading metrics.
To accommodate exchanges located in different regions, the trading system is designed as a geo-distributed architecture. The central system is deployed on the main server, functioning as the hub for collecting and processing market information. Near each exchange server, smaller gateways are strategically positioned to intercept data directly from the exchanges. The chosen protocol for data transfer is UTP, known for its high-speed capabilities.
This module allows the central system to gather real-time data from multiple exchanges. The gathered data includes quotes, the current state of order books, fundings, and other information that provides our client with a comprehensive market overview. The system applies machine learning approaches to identify anomalies in the market, which empowers the client to make trading decisions based on the understanding of market dynamics.
The order management module facilitates the efficient handling and monitoring of the order book. The system allows our client to keep track of order status in real-time, handling numerous orders simultaneously.
This module includes order creation, order sending, and continuous monitoring of the execution status. By offering immediate order placement, the system enables traders to capitalize on advantageous price levels swiftly.
Furthermore, it provides instant order status updates, ensuring traders have complete visibility into the execution process. Our client can monitor the progress of the orders, following the progress of full or partial order execution. There are also features such as order level approvals, where traders have the option to approve orders based on specific predefined criteria.
The positions manager provides traders with real-time visibility into their current trades, balance control, and a comprehensive overview of their remaining funds. This tool allows traders to monitor their portfolios and assess their exposure to different assets. The module provides additional details, such as the average purchase price, current market value, and unrealized gains or losses associated with each position. This module also interacts with the risk manager to control trading operations and limits.
The cryptocurrency trading platform provides traders with full control over orders, purchases, and risk assessment. By incorporating risk parameters, this module ensures that orders are executed within acceptable price ranges. The tool’s primary function is to monitor and control order execution in relation to real-time market prices based on ML analysis. A set of algorithms ensures that purchase prices remain within predefined limits. By comparing the executed price to the prevailing market price, the module helps traders avoid significant deviations that could impact profitability. Additionally, traders can set specific loss tolerance levels tailored to their risk preferences and trading strategies. This function allows for the establishment of predefined loss limits based on asset types and trading operations. The module provides real-time monitoring of Profit and Loss (PnL) positions and the current profitability status to adjust their strategies accordingly. The risk management module also offers advanced risk assessment tools, allowing traders to evaluate the potential risk associated with specific trades or portfolio positions. By analyzing factors such as asset volatility, historical price movements, and correlation analysis, traders can gain deeper insights into their risk exposure and adjust their risk management accordingly.
The module for trading strategies is responsible for implementing and executing automated trading algorithms based on predefined logic and market conditions. This module combines machine learning techniques, particularly boosting algorithms, with the client’s specific trading plan to generate actionable insights and execute trades in real time.
At the core of the module is the strategy itself, represented as a separate class, which encapsulates the trading logic and defines the actions to be taken under various market scenarios. By working with relevant datasets using machine learning, the module identifies and extracts data features to train models that automatically implement the strategies based on current conditions.
The process begins with training the ML models using the selected datasets. These models analyze and process market info, including trading volumes, to detect anomalies and determine optimal entry or exit points for specific assets. Using boosting algorithms, which provide enhanced accuracy, the models generate predictions for asset prices within the shortest possible time intervals, such as milliseconds.
The ML models communicate with the trading system’s backend, where the resulting predictions are stored in a database for further analysis and decision-making. As market data arrives from the exchanges, the models evaluate the conditions against predefined requests and criteria. Based on these evaluations, the models generate predictions that inform buying or selling decisions.
The models continuously learn and adapt to market patterns, improving their predictive capabilities over time. This enables the system to capture price discrepancies across different exchanges promptly, identifying opportunities to sell at higher prices or buy at lower prices.
The module’s architecture is designed to support multiple exchanges offering similar trading tools. Its primary objective is to capitalize on market fluctuations by swiftly identifying favorable trading opportunities. By incorporating trading volume data and ML-driven anomaly detection, the tool enhances the probability of executing trades.
Throughout the development process, Innowise Group as the trading software development company followed a structured and efficient process to ensure successful collaboration with the client. The project workflow encompassed three main stages:
Our team is actively expanding the project by integrating new data collection exchanges. Our goal is to make the project highly competitive and unique in the market. To achieve this, we are in the process of rewriting the codebase in C++ to enhance its speed and performance further. In addition, we are considering rewriting frequently used connectivity libraries from scratch to speed up the system’s performance.
Our development of the custom quantitative trading platform yielded significant improvements for the client. The system’s ultra-fast infrastructure reduced information processing delays from an average of 2-3 seconds to 34 milliseconds, resulting in a remarkable speed improvement of approximately 97%. By leveraging machine learning techniques, the system improved the client’s trading strategies, leading to an increase in profitability. The system’s ability to capture arbitrage opportunities and react swiftly to market movements allowed the client to outperform competitors, while the risk management tools effectively manage orders and purchases, leading to a reduction in potential losses.
Innowise Group developed a user-friendly API that simplifies strategy development and testing. Our client no longer needs to invest significant time in working with third-party resources, as everything can now be done within our unified system. Additionally, the API we created provides clear and comprehensive metrics for each strategy, enabling our client to assess its suitability for their risk profile easily.
We have also significantly accelerated the development of Gateways for exchanges. By transitioning from a monolithic architecture to microservices, we have reduced the time required for Gateway development. Our team is currently dedicated to enhancing the quantitative trading platform, aiming to establish it as a distinctive and unparalleled tool for online crypto trading in the market.
faster trading information processing
milliseconds market response time
Having received and processed your request, we will get back to you shortly to detail your project needs and sign an NDA to ensure the confidentiality of information.
After examining requirements, our analysts and developers devise a project proposal with the scope of works, team size, time, and cost estimates.
We arrange a meeting with you to discuss the offer and come to an agreement.
We sign a contract and start working on your project as quickly as possible.