How Quant Research Teams Are Organised

Ask ten quant researchers at ten different firms what their job looks like day to day, and you'll get answers that barely resemble each other — not because the underlying work differs that much, but because the organizational shape they sit inside differs enormously. A researcher's title tells you almost nothing about their actual role until you know how the surrounding team is structured.

The idea

At one extreme is the fully embedded researcher: one or two people attached directly to a trading desk or portfolio manager, working exclusively on that book's problems, with fast feedback loops and deep context on one strategy but little exposure to research happening elsewhere in the firm. At the other extreme is a large centralized research division, organized more like an academic department, that builds infrastructure and signals meant to be reused across many books, with researchers who may never speak to the PMs who eventually trade their work. Most firms sit somewhere between these poles, and the specific point on that spectrum shapes everything from how ideas get greenlit to how credit gets assigned.

A second axis, independent of centralization, is specialization versus generalism. Some teams organize by asset class (an equities research group, a rates research group), some by research function (a signal-generation team, a portfolio-construction team, a data-engineering team that supports both), and some by neither, expecting each researcher to own an idea end to end from raw data to live capital. Specialized structures tend to produce deeper expertise per person and clearer accountability, at the cost of more handoffs and more coordination overhead between the specialists. Generalist structures move faster on any single idea but scale worse, because each researcher's knowledge is harder to transfer when they leave.

A concrete example

Consider two hypothetical firms researching the same idea — a new event-driven signal around merger announcements. At an embedded-desk shop, one researcher owns the entire pipeline: sourcing the M&A data, building the signal, backtesting it, and pitching it directly to the PM they sit next to, with a decision inside two weeks. At a centralized-research shop, the same idea might pass through a data-acquisition team to source the feed, a signal-research team to build and validate it, a portfolio-construction team to decide how it interacts with existing positions, and a review committee before any capital is allocated — slower, but with more scrutiny at each stage and a signal that, if it works, is immediately available to every book in the firm rather than just one.

What this means in practice

Neither shape is objectively better; they trade speed and single-strategy focus against reuse and scrutiny, and the right shape depends on firm size, strategy count, and how correlated the firm's various books are. For a researcher choosing where to work, this is worth investigating directly in interviews — "how is research organized here" reveals more about day-to-day life than almost any other question, because it determines whether you'll spend your time deep on one problem or coordinating across several.