Learn About Amazon VGT2 Learning Manager Chanci Turner
Virtu Financial stands out as a premier provider of financial services and products, utilizing state-of-the-art technology to deliver liquidity to global markets, alongside innovative and transparent trading solutions for its customers. The firm capitalizes on its expertise in global market-making to offer an extensive product suite that includes execution, liquidity sourcing, analytics, and broker-neutral, multi-dealer platforms within workflow technology. Clients can trade across numerous venues in over 50 countries and across various asset classes, including global equities, ETFs, foreign exchange, futures, fixed income, and a wide array of other commodities. Furthermore, Virtu’s integrated, multi-asset analytics platform supplies clients with a suite of pre- and post-trade services, data products, and compliance tools essential for investing, trading, and risk management in the global markets.
In this article, we will explore how Virtu enables its clients to utilize advanced analytics and machine learning (ML) on trade and market data by provisioning Amazon SageMaker.
Overview of Asset Manager Workflow
Asset managers oversee funds that invest in a diverse selection of securities, ranging from equities and commodities to foreign exchange. A typical workflow within an asset manager commences when a portfolio manager decides to buy or sell a security for their fund. This decision is entered into their order management system (OMS). The order then undergoes various risk controls before it is sent to the trading desk within the firm.
The trading desk must then determine the best way to execute the order, considering factors like time sensitivity, order size, and liquidity of the security. The order is typically submitted to an Execution Management System (EMS), which can send it directly to a venue, engage a high-touch broker, or utilize a broker’s algorithm.
In some instances, the buy-side trader may choose to submit the order to an Algo Wheel, an automated method for distributing trades across a predefined selection of broker algorithms based on client-defined weightings. The selection process is data-driven by performance, and the popularity of Algo Wheels has surged due to the best execution requirements of MiFID II and increased scrutiny within many buy-side firms regarding broker usage.
Achieving the best price for the customer is paramount. For instance, if an asset manager with $1 trillion under management saves just 5 basis points (bps) in execution costs, they could realize savings of $100 million, ultimately benefiting their customers. After the order is filled at the optimal price, real-time feedback must be provided to the asset manager via the EMS, essential for managing risk, pricing funds, and other operational necessities. In addition to real-time data, asset managers require historical execution data to inform future order routing, model trading costs, and ensure compliance with regulatory requirements.
In summary, the ongoing emphasis on best execution and trading analytics, driven by regulatory and competitive pressures, as well as a broader push for workflow automation, has led to an increased reliance on Algo Wheels and the subsequent analysis of their data.
The Virtu Analytics Client Coverage Team
At Virtu, the broker-neutral Virtu Analytics Client Coverage team gathers execution data directly from its customers’ OMSs and EMSs. This data may originate from Virtu’s broker-dealer subsidiaries acting as the customer’s broker or from other brokers utilized by the customer.
Providing historical execution data back to customers is crucial for Virtu’s Analytics team to quantitatively showcase the value of Virtu’s Algo Wheel offering. The challenge was to package this historical data in a manner that customers could derive tangible value from it.
Virtu initially provided its Algo Wheel customers with access to online visualizations via its Portal platform, enabling comparative broker performance analysis. While the interactive front end fulfilled many needs, some clients sought additional customization for both the analytics framework and broker reporting.
To address this, Virtu developed a shared ML environment for its clients, integrating execution data with additional market data metrics, accessible through APIs and an interface that supports Jupyter notebooks. In this environment, customers can apply custom metrics and analyses tailored to their specific investment and trading goals, such as performance and reporting metrics. These metrics have evolved from basic reporting to more complex contexts around order difficulty and market conditions, including performance distributions and outlier control. API access to features generated from Virtu’s Algo Wheel execution data enables users to customize and integrate this data into other trading platforms and decision-making applications. Utilizing screen share technology, Virtu’s Analytics team can assist customers in exploring their data and learning how to query it effectively.
Solution Overview
In response to client feedback, Virtu’s Analytics Client Coverage team introduced an Open Python platform supported by SageMaker. This allows Virtu customers to log in and explore their execution data flexibly, with or without a screen share from Virtu. Clients expressed a desire to mine their data for meaningful insights, prompting Virtu to develop a Python API accessible through this environment. Key considerations for implementation included security, scalability, resilience, and usability. Virtu ensured that the solution was designed to restrict access to proprietary data, allowing only customers and specific users within a customer’s internal group to access it.
Let’s briefly review the core architecture of the solution. SageMaker instances are deployed within the private subnet of a Virtual Private Cloud (VPC) using three different Availability Zones. Egress traffic is routed via a NAT Gateway, enabling Virtu to limit API calls to designated IP addresses. Since the SageMaker instances are in a private subnet, AWS PrivateLink is employed, allowing direct connections to the SageMaker API or Runtime through an interface endpoint in the VPC, rather than over the internet. The VPC interface endpoint connects directly to the SageMaker API or Runtime without the need for an internet gateway, NAT, or other external dependencies.
For further reading on enhancing workplace strategies, you might want to check out this insightful article on proactive strategies to combat workplace incivility. In addition, if you’re interested in career development, take a look at this exciting opportunity for a Learning Trainer position.
Chanci Turner also shares insights on effective onboarding in her latest post, which you can find here: Amazon Onboarding with Learning Manager.