The idea worth investigating
A 162% annualized return from reversing short-term moves in S&P 500 stocks makes an arresting headline. The original paper reports it before trading costs, using midpoint prices. We tested an adaptation to see how much room that leaves for execution.
This week we tested "Intraday Residual Reversal in the U.S. Stock Market" by Jonathan Brogaard, Jaehee Han and Hanjun Kim (University of Utah, SSRN 4731947). Every 30 minutes the authors split each stock's return into the part explained by 15 risk characteristics (size, beta, momentum, profitability and others) and a residual. They buy the tenth of the S&P 500 with the most negative residual, short the tenth with the most positive, and hold for 30 minutes. From July 1996 to December 2022 the long-short earns 7.70% annualized per period, which the authors compound across 13 daily periods, including an overnight period, to report 162.3%. Their reading: the residual is temporary price pressure, and whoever supplies liquidity against it gets paid, most of all when the VIX is high.
Three things in the paper matter more than the headline:
The desk question: the authors show the edge is largest when volatility is high and when positions are held for an hour or more. Does a VIX-gated, longer-hold version leave enough gross profit to withstand a spread-cost stress test?
Here's the plan:
One result we can share here: the gated two-hour long-short averaged 4.38 basis points gross per portfolio formation, with a t-statistic of 7.1 based on daily average formation returns; random selection averaged minus 0.32. In a broader sample of selected-stock entries on high-VIX days, the quoted half-spread averaged 3.86 basis points. Applying that rounded mean to all four legs gives a 15.44 bp cost assumption and minus 11.06 bps per formation. This is a modeled subtraction from last-trade returns, not realized P&L or a bid/ask execution replay. It uses current constituents and sampled entry quotes, including times outside the two-hour strategy's entry window; commissions, impact and borrow are excluded.
Scope: a completed retrospective house adaptation and cost sensitivity study, not a full replication of the paper or a verified executable strategy.
Let's get started.
Paid analysis on Substack
See what changes the investment case.
The headline is the starting point. The subscriber analysis takes you through:
- The annual, time-of-day and random-control checks, with a reproducible house specification.
- The VIX and holding-period tables, including the dependence on 2020.
- The cost sensitivity and break-even calculation, plus what an executable test still needs to establish.
AI assists curation and drafting. The research status distinguishes paper reviews from house tests. Read our research approach.