A signal that works is small, short-lived and expensive to find

The idea

Zuckerman's account of Renaissance is the best available to someone who is not a quant, and its central lesson is about the shape of an edge. The signals that worked were tiny, each explaining a sliver of price movement, and they only paid because they were applied across enormous numbers of positions with leverage. They decayed as others found them, so the work never stopped. And finding them cost years of effort by unusually talented people, which is the barrier that kept the returns from being competed away. Read against the scouting story in Moneyball, it is the same argument taken to its limit.

Expect an edge to be small, to fade, and to cost more to find than to run; a large, durable, obvious one is probably not an edge.

How to use it

Asked about a strategy you have read about, say why it is likely to have decayed since publication and what would have to be true for it to persist. That is the question a quant desk is actually asking.