Quant Forensics: Trade Timing: Ten Years On... Revisiting Fast-Burning Signals for Slower Quant Strategies in Global Equities

研报英文原文证据摘录

Quant Forensics: Trade Timing: Ten Years On... Revisiting Fast-Burning Signals for Slower Quant Strategies in Global Equities

Robert Smith, PhD AC Global Markets Strategy

(61-2) 9003-8808 06 July 2026 J P M O R G A N

robert.z.smith@jpmorgan.com

Trade Timing: Concept & Rationale

Strategies with high turnover are hard to take in anything but very small doses. We often

see great back-tests, only to find that the gross returns are eroded by implementation

costs. Short-term reversion is a classic example. The performance is quite robust in most

global markets, but with a one-way turnover of 80–90% per month, it is prohibitive for

most institutional investors once realistic costs are included.

The standard response has been to include these fast-burn factors sparingly within a

slow composite: a small cross-sectional weight, in the hope of incremental alpha

without dragging turnover too high. The trade-off is rarely satisfying. The more weight

the fast factor gets, the more its turnover bleeds into the composite. The less weight it

gets, the less of its alpha survives.

Trade timing uses these signals differently. The simplest way to put it: keep buying

what a slow strategy wants to buy, but only when a faster signal says it's the right time

to buy it.

Two models run side by side. The slow model produces the what, a wishlist of names

we believe in for the next several months. The fast model produces the when and the

entry timing for each name on that list. The slow model still drives the investment

thesis. The fast model only relaxes the rigid, fixed rebalance date. If the fast signal never

agrees within the holding window, the name simply does not enter the book.

Because the fast signal never enters the book, its turnover does not show up in our

trading costs.