Atlantic House received the Model Master award. Mark Greenwood, Deputy Chief Investment Officer at the asset manager, described his firm’s efforts to use BQuant Enterprise’s data-driven tools to analyze economists’ inflation estimates. Using an array of statistical tools, including L-estimators, and random forest analysis, Atlantic House’s team was able to generate inflation estimates based on forecasts from economists with the best record for accurate inflation predictions. Greenwood calculates that for every one basis point improvement in the accuracy of economists’ mean inflation estimates, his firm can boost returns on its UK macro strategy by approximately 0.1 Sharpe points.
J.P. Morgan was named the Visual Insighter. Daniel Thiel, Executive Director, Quantitative Research, Equities, described how his firm built an application using BQuant Enterprise that employs textual analytics on a dataset based on Bloomberg News to flag earnings warnings, or changes in corporate profit forecasts for J.P. Morgan’s equity traders and salespeople. These announcements can move markets sharply, like when retailers reported a spate of downward revisions set off by new tariffs this spring. Thiel’s team automated this process using the Python data science tools available within BQuant Enterprise. The application automatically pulls the latest, most accurate Bloomberg News headlines regarding profit warnings, providing a summary and text preview on the application user’s Terminal Launchpad. It can be used to develop both company-specific or industry-wide perspectives, or to highlight earnings trends at related companies. In addition, it offers month-on-month data to highlight longer-term trends.
The event concluded a four-month intensive development process in which teams from all over the world used BQuant Enterprise to explore new use cases for the programmatic analytics solution, with the support of the BQuant Enterprise team. “When we first started planning Code Crunch, we set out to do something ambitious. We want to create a space for experimentation, shared learning and best practices across the BQuant community,” said Vuong.
Enhancing factor models with machine learning techniques
In addition to highlighting client work using BQuant Enterprise, the event showcased the BQuant Enterprise team’s own capabilities. Bloomberg’s own Anish Popat, a market specialist, and Ricard Radomski, a Desktop Build Group analyst, explained how an example framework they built incorporates machine-learning-based signals into a multifaceted investment analysis.
“These signals can capture complex and nonlinear relationships that standard models often miss,” Popat explained. “By blending the traditional insights with modern techniques, our goal is to refine the signal, and ultimately, a better, more refined signal should lead to better decisions, whether you’re building a portfolio or screening for stocks.”
Code Crunch is just the beginning
On their own, the projects created through Code Crunch are impressive. Even more important is the process of collaborative innovation generated by this event.
“Whether the client teams were alpha innovators, model masters, or data storytellers, every project brought unique value to the table,” said Dupire. “Code Crunch was about pushing the limits of BQuant Enterprise, and these are the kinds of investment analyses that we want to empower our clients to do every day.”
