4 Integrated Steps to Get from Research to Deployment Faster

By: Mark Higgins and Mark Ayzenberg

Benefits of a Batteries Included Approach to Quant Research and Data Science

As a quant researcher or data scientist, you want the quickest way to get your ideas into algorithms and running in production, so that they can help the firm’s trades or risk managers.

A common problem in this space is that R&D work requires production-quality data—and cobbling that data together in a research environment can be tricky and time-consuming. Then once you find something that works, it’s often inefficient and bureaucratic to get that new functionality into production where it can help your business make money.

There is a better way. Beacon Platform provides a batteries included approach—integrated workflows that enable you to move seamlessly from prototype to production, testing, reviewing, and refining as you go. The result is huge savings in development effort, better reproducibility, and faster time to market.

Step 1: Research

Starting with an idea, the first step is building the model or algorithm and researching how it works against real data. Beacon is designed to be both your research environment and the production system used by the business—so the research environment is automatically connected to your production data, ready to use for new ideas. Data comes in many different shapes from different sources, so Beacon is an open platform that easily works with existing data sources and controls. Beacon’s data and analytics are also directly accessible from Jupyter Notebooks, a common tool for data science research. Quickly sketch out, test, and refine algorithms and analytical strategies in Python. Integrated plotting tools facilitate the analysis with quick data visualizations. All of this is readily shared with other members of the team and does not require advanced knowledge of the underlying infrastructure.

Step 2: Prototype

Whether you are working on a large data science project or small changes to an existing model or algorithm, eventually the time comes to begin the move from research to production. Without an integrated environment this production deployment traditionally requires a separate technology team to reimplement models in the production environment, which can be bureaucratic and time-consuming. Since Beacon is both a research and production environment, there is no change of tools. The code from the Jupyter Notebook is simply restructured as a feature branch and pushed into the review workflow. Beacon abstracts the underlying complexity, giving you the power of the Git distributed version control system, enterprise controls, review and testing requirements, and production workflows without having to be an expert. Most other development products expect you to configure and maintain this yourself.

Step 3: Scaling

At some point during the development cycle you want to run your code against large datasets or multiple scenarios, requiring a significant amount of compute and storage. Cloud infrastructure is perfect for this, providing compute pools that run huge scenarios quickly and inexpensively, but can be daunting to set up and learn. This is yet another integrated part of Beacon Platform, abstracting the infrastructure, process, and control tools for multiple cloud providers. Beacon’s compute grid and batch job schedulers give you ready access to the resources you need, automating repetitive and complex tasks so that you can focus on the results. The job scheduler is directly integrated with the development environment and cloud infrastructure for seamless and secure workflow automation. Dashboards provide full control and transparency, simplifying operations and reducing administrative costs.

Step 4: Deployment

Review and testing have gone well, and faster than expected, so you are now ready to deploy the new or updated code into production. Since Beacon is both a development and production environment, this step is just another part of the integrated workflow. All source code is categorized with appropriate controls for the release process. Your analytical report is now available for traders and portfolio managers to run against their portfolios. Beacon includes a full deal and portfolio model to achieve consistency in pre-trade analytics, pricing, and post-trade risk management.

With Beacon, Batteries Are Included

Beacon is the only vendor with a collaborative platform that includes everything you need to streamline the path from research to production, so that you can quickly make money from your quantitative and data science innovations. With all of the “batteries” included, you can build and test research signals faster, and easily incorporate whatever data you need. Included applications, data visualizations, business intelligence tools, and the full underlying source code, let you focus on your areas of expertise and competitive edge, not redoing basic functionality. Finally, full enterprise controls for code management, collaboration, and governance, plus an integrated application development framework, makes the process of moving your ideas from inception to deployment secure and efficient.

To learn more about Beacon Platform or request a demo, contact info@beacon.io