Research Idea Intake and Triage

A research team hears far more candidate ideas than it can ever build and backtest properly — a paper someone read, a pattern a colleague noticed, a strategy a competitor is rumored to be running. Without a structured process, whichever idea was pitched most recently or most persuasively tends to get the attention, regardless of whether it's actually promising. Intake and triage is the discipline of capturing every idea in one place and running it through a lightweight, consistent first screen before anyone spends real backtesting effort on it.

A typical intake process logs each idea with a short written description, the economic or behavioral rationale for why it should work, and any prior evidence — an academic paper, a related strategy already on the book, informal observation. Triage then applies a small set of cheap screening questions before committing to a full build: is the rationale specific enough to be tested, or is it vague enough to fit any result after the fact? Is there capacity in the relevant market for the strategy to matter at the size the fund would trade it? Does it overlap heavily with a strategy the fund is already running, such that it likely shares the same risk rather than diversifying it? Ideas that fail this cheap screen get archived rather than built, freeing research capacity for ideas that pass.

A concrete example: a research team logs fifteen new ideas in a quarter. A quick triage pass eliminates four for lacking a specific testable rationale ("this factor should work because markets are inefficient" without saying why or where), two for targeting a market too illiquid to matter at fund size, and three for being close variants of a strategy already running. That leaves six ideas worth a proper backtest — a shortlist a team can actually give serious attention to, rather than doing a rushed, shallow pass across fifteen at once.

Triage is not meant to be a rigorous filter — it's a fast, cheap first pass specifically so the expensive, careful work of backtesting and validation is reserved for ideas that clear a low bar, rather than spent thinly across everything that comes in the door.