End of Day Reversal Trading Strategy (Day Trading)

Why do intraday losers pop in the final 30 minutes of the trading session? This article shows you the abnormal returns obtained from buying intraday losers and what might be the reason for this anomaly.

Key Takeaways

• Intraday losers systematically rebound in the last 30 minutes, while winners tread water.
• Retail contrarian buying and short‑covering pressure are the main culprits.
• A simple, time‑boxed long‑short strategy exploiting the pattern produced >20 % annualized alpha (gross) from 1993‑2019.
• Real‑world implementation requires disciplined cost control and size limits.
• Reference: Baltussen, G., Da, Z., & Soebhag, A. (2024). End‑of‑Day Reversal.
• More day trading articles and strategies.

What Is the End‑of‑Day Reversal?

The end‑of‑day reversal (EODR) is a striking intraday return pattern uncovered by Baltussen, Da & Soebhag (2024).

After ranking all stocks by their return from the previous close up to 3:30 p.m. Eastern Time (the rest‑of‑day‑3 or ROD3 window), the researchers find that the biggest intraday losers reliably outperform the biggest winners in the last 30 minutes of the session (3:30–4:00 p.m.).

In other words, today’s laggards stage a mini‑comeback just before the closing bell, while early‑day winners tend to stall or slip.

Key Findings at a Glance

• Economically large and statistically strong. A simple long‑short portfolio that buys the bottom ROD3 decile and sells the top decile earns an average 0.24 % return every day in the final half‑hour.
• Distinct from intraday momentum. The pattern is cross‑sectional (stock‑specific) rather than a market‑wide drift. It is the mirror image of the well‑known morning‑to‑afternoon intraday momentum documented at the index level.
• Driven mainly by losers. Positive price pressure on intraday losers, not negative pressure on winners, accounts for most of the reversal.
• Robust. The effect survives controls for size, liquidity, volatility, relative bid–ask spreads, and gamma‑hedging demand.

Data Set and Methodology (1993 – 2019)

The authors analyze common stocks (CRSP share codes 10 & 11, price > $5, above the 10th NYSE size percentile) listed on NYSE, NASDAQ, and AMEX from January 1993 through December 2019.

Trades and quotes come from the TAQ database, while daily returns and firm characteristics come from CRSP/Compustat. Each trading day is partitioned into five windows:

1. Overnight (ON): previous close → open
2. First Half‑Hour (FH): open → 10:00 a.m.
3. Mid‑day (M): 10:00 a.m. → 3:00 p.m.
4. Second‑to‑Last Half‑Hour (SLH): 3:00 p.m. → 3:30 p.m.
5. Last Half‑Hour (LH): 3:30 p.m. → 4:00 p.m.

The sum of windows 1–3 is ROD3 – the key predictor of last‑half‑hour returns.

How the End‑of‑Day Reversal Trading Strategy Works

At 3:30 p.m., rank all eligible stocks by ROD3 return (from the previous close to 3:00 p.m.). A hypothetical spread portfolio that goes long the worst-performing decile and tracks the return difference with the best-performing decile from 3:30–4:00 p.m. shows strong reversal. (Note: the study presents this as a return differential, not a traded long-short portfolio involving actual short positions.)

A potential trading strategy could be:

1. At 3:30 p.m. rank all eligible stocks by ROD3 return (previous close → 3:00 p.m.).
2. Go long the bottom 10 % (intraday losers) and short the top 10 % (intraday winners).
3. Hold for 30 minutes until the 4:00 p.m. close. Positions are closed before the overnight session begins.

Backtest highlight: Over 1993‑2019, the dollar‑neutral decile‑spread portfolio compounds to more than 10× the return of the broad market (Ken French’s CRSP market factor) held continuously, despite being invested only 0.5 hours per day.

This ”end-of-day reversal” pattern is economically and statistically highly significant, and holds across almost every 3-year rolling window, methodologies, and various subsamples of stocks, including the largest, most traded, or most liquid stocks.

The equity curve of the strategy looks like this:

Why Does the Reversal Happen? Two Main Reasons

1. Attention‑Induced Retail Buying (Buy the Dip)

Retail investors tend to place attention‑driven market orders near the close (after work hours, push‑notification prompts, media recaps, etc.). Retail investors and traders tend to buy and sell at the open and at the close. This is a classical buy the dip strategy.

Intraday losers gain visibility by showing up on “biggest decliners” lists, attracting contrarian bargain hunters. TAQ order classifications reveal that retail‑dominated brokerages supply a disproportionate share of the last‑minute buy‑pressure in these names.

2. Short‑Sellers’ End‑of‑Day Risk Management

Professional short‑sellers face heightened inventory and margin requirements overnight. Many, therefore, reduce exposure in the final minutes, especially in stocks that have already fallen intraday (and thus show elevated volatility).

Covering shorts mechanically lifts the prices of intraday losers.

Other (Less Pivotal) Explanations Considered and Rejected

• Gamma‑hedging by options market‑makers – no systematic link to EODR after controlling for option open interest.
• Liquidity imbalances – bid/ask spread effects are too small to explain the magnitude.
• Arbitrageurs unwinding positions – accounts for only a fraction of the return gap.
• Less short-selling in the final 30 minutes because short sellers don’t want to take on overnight risk.

Relation to Intraday Momentum and Closing‑Auction Effects

While index‑level studies document positive intraday momentum (morning winners keep winning until the close), EODR shows the opposite at the individual stock level.

The two patterns coexist because the index effect is about aggregate order‑flow persistence, whereas the cross‑sectional effect reflects differential flows into specific stocks.

Practical Implications for Traders and Portfolio Managers

• Low capital tie‑up. The strategy is active for just 5 % of the trading day, freeing capital for other uses.
• Execution matters. Slippage and exchange‑closing‑auction rules can erode a sizable chunk of the edge; algorithmic slicing or auction participation may help.
• Scalability limits. Because returns concentrate in small‑ and mid‑caps, capacity is modest.
• Diversification tool. Returns exhibit low correlation with daily momentum, value, and other well‑known factors.

Limitations and Risks

• Transaction costs. Even with TAQ‑based cost estimates, net performance falls by ~40 %.
• Regime shifts. The effect weakened modestly after 2010 as HFT participation rose.
• Crowding. Public knowledge of the anomaly could compress future returns.

Reference

Baltussen, G., Da, Z., & Soebhag, A. (2024). End‑of‑Day Reversal. Working paper. SSRN ID 5039009.