The Last 30 Minutes of Trading: What Research Says About Intraday Momentum

Traders often describe the final part of the trading session as unpredictable. Volume rises, institutional orders enter the market and prices can move rapidly before the closing bell. However, academic research suggests that the final 30 minutes may not be entirely random.

In the working paper Hedging Demand and Market Intraday Momentum, Guido Baltussen, Zhi Da, Sten Lammers and Martin Martens examine decades of intraday futures data. Their main finding is straightforward:

In practical terms, markets that were up going into the final half-hour were, on average, more likely to continue rising into the close. Markets that were down were more likely to continue falling.

The effect appeared across equity-index, government-bond, commodity and currency futures. The researchers connect it partly to hedging activity by option market makers and leveraged exchange-traded funds. This does not mean the final half-hour is easy to predict; the relationship is probabilistic rather than deterministic. But it is a clear example of how institutional market mechanics can create recurring intraday price patterns.

What did the researchers study?

Baltussen and his co-authors studied intraday returns for more than 60 futures markets across four major asset classes: equity indexes, government bonds, commodities and currencies.

The sample covered approximately 45 years, from December 1974 until May 2020. It included 17 developed-market equity-index futures, 16 bond futures, 21 commodity futures and eight currency futures.

The study used one-minute market data and divided each trading day into several periods: the overnight session, the first 30 minutes after the open, the middle of the session, the second-to-last 30 minutes and the final 30 minutes before the close.

The researchers called the period from the previous market close until 30 minutes before the current close the rest of the day, or ROD. For US equity-index futures, that roughly means the return from the previous 4:00 p.m. cash close until 3:30 p.m. on the current day. They then asked whether that return could predict the market's direction between 3:30 p.m. and 4:00 p.m.

The central finding: intraday direction often continues into the close

The researchers found a positive relationship between the rest-of-day return and the final-half-hour return. A positive rest-of-day return was associated with a positive expected return during the final 30 minutes, and a negative rest-of-day return with a negative expected return.

The return over the full rest of the day was a better predictor than the overnight and opening return alone. In equity futures, the pooled out-of-sample R² was 2.88%, and the result was positive and statistically significant for 14 of the 17 individual equity-index futures studied.

An R² of 2.88% may sound small. In financial return forecasting, however, even a low explanatory value can be meaningful, because short-term returns contain substantial noise. It is still essential to read the number correctly: the predictor did not explain most final-half-hour price movements. It simply improved predictions relative to using the historical average return.

The findings for Nasdaq-100 futures

The study is particularly relevant to Nasdaq futures traders because Nasdaq-100 futures were included as an individual market. The NQ sample ran from April 1996 until May 2020 and used the regular US session of 9:30 a.m. to 4:00 p.m. Eastern Time. For NQ, the rest-of-day predictor produced:

The values in the paper's table are multiplied by 100, so the underlying regression coefficient is approximately 0.0636. A simplified reading is that a 1% NQ return from the previous close until 3:30 p.m. was associated with an average final-half-hour return of roughly 0.0636% in the same direction. That is an average historical relationship; it does not imply NQ continued in the same direction every day.

The out-of-sample result matters most. Out-of-sample testing evaluates whether a model can predict observations it was not fitted on, which reduces the risk that the finding merely reflects overfitting within the original data.

How economically significant was the effect?

The authors also tested a simple market-timing strategy: take a long position during the final half-hour when the rest-of-day return was positive, and a short position when it was negative, holding only during the final 30 minutes. For equal-weighted portfolios within each asset class, the historical results were:

The success rates are instructive: the strategy was correct on only about 53% to 56% of trading days. The historical performance did not come from exceptionally accurate predictions but from a relatively small directional advantage repeated across many markets and many days. The headline Sharpe ratios should also be read cautiously, because they represent diversified portfolios of multiple futures contracts, not NQ-only results.

Why might the effect exist?

The authors argue that hedging demand is an important driver of final-half-hour momentum, and their explanation centres on gamma exposure. (For a full primer on how gamma and dealer hedging work, see our guide on how to read a gamma exposure (GEX) chart.)

What is gamma?

Gamma measures how quickly an option's delta changes when the price of the underlying asset changes. Delta estimates how much the option price changes for a given move in the underlying; gamma determines how quickly that delta exposure itself changes as the market moves. An option market maker may hedge a position by buying or selling the underlying, and as the price changes, the required hedge must be adjusted too.

What happens when market makers are short gamma?

A short-gamma position creates procyclical hedging behaviour. When the market rises, a short-gamma market maker may need to buy more of the underlying; when it falls, the same maker may need to sell. The hedging occurs in the same direction as the existing move:

This can amplify price changes and create momentum. By contrast, long-gamma hedging normally occurs against the direction of the move, a long-gamma participant may sell after the market rises and buy after it falls, contributing to mean reversion rather than continuation.

The evidence for the gamma explanation

The researchers created an estimate of the net gamma exposure of S&P 500 option market makers, then separated trading days into negative net gamma days and positive net gamma days.

On negative-gamma days, the relationship between the rest-of-day return and the final-half-hour return was strong and statistically significant. On positive-gamma days, there was no significant intraday momentum. For negative-gamma days, the regression coefficient was 6.63, with a t-statistic of 4.78 and an R² of 3.58%; on positive-gamma days the coefficient was much smaller and statistically insignificant. The effect also became stronger as estimated short-gamma exposure became more negative.

This supports the idea that forced or systematic hedging flows contribute to late-day momentum. Note, though, that the gamma analysis was conducted specifically on the S&P 500. The paper documents intraday momentum in NQ, but it does not provide the same Nasdaq-specific gamma test.

Leveraged ETFs may also reinforce the move

Leveraged and inverse ETFs try to deliver a multiple of an underlying index's daily return. Because their leverage target resets daily, they often need to rebalance near the close. After the market rises, a leveraged-long ETF may need to increase exposure to restore its target leverage; after a fall, it may need to reduce exposure. Inverse ETFs also create rebalancing demand that tends to occur in the same direction as the underlying's daily move.

The paper finds that markets with greater leveraged-ETF activity displayed stronger intraday momentum, and that estimated hedging demand from these products added predictive information for the final-half-hour return. This helps explain why the final part of the session can behave differently from the middle of the day.

Why does the effect concentrate near the close?

Several practical reasons may cause hedging activity to cluster near the closing bell. First, liquidity is generally high near the close, so more depth and narrower spreads make it easier for large participants to execute sizeable orders. Second, market makers may want to reduce overnight risk, since positions held after the close can involve extra uncertainty, margin and capital costs. Third, many products are benchmarked or settled using official closing prices, so leveraged ETFs, index options and some variance-related products have structural reasons to complete their hedges near that point.

Tellingly, the authors found that the predictive relationship ended at the 4:00 p.m. cash close. Although S&P 500 futures continued trading actively between 4:00 and 4:15 p.m., the rest-of-day return did not predict the futures return during that later interval. This supports the hedging explanation: the effect was tied to the closing time of the underlying market and related products, rather than being a general end-of-session tendency.

The price pressure later reversed

The paper also examined what happened after the final-half-hour momentum occurred. The move tended to reverse over the following days: within roughly one day for currencies, two days for equity and commodity futures, and three days for government bonds.

This reversal matters, because it suggests the late-day move reflects temporary price pressure. If the move were entirely caused by new fundamental information, its price impact would tend to remain. A later reversal is more consistent with mechanical hedging orders temporarily pushing prices away from their previous level.

What value does the research have for intraday analysis?

1. Time of day matters

A price move does not necessarily carry the same meaning at every moment of the session. The final 30 minutes contain specific institutional flows tied to closing auctions, portfolio rebalancing, options hedging and leveraged products.

2. Market direction can interact with market structure

The day's existing direction may affect the hedging demand that enters near the close. A large positive or negative daily return changes the delta exposure of option positions and the rebalancing needs of leveraged products, so the earlier move can influence later order flow.

3. Gamma is a conditioning variable, not a complete prediction model

The results suggest negative gamma strengthens intraday continuation, but gamma estimates do not reveal every participant's actual position. The paper's gamma measure relies on assumptions about who holds calls and puts and how those positions are hedged, and it excludes certain over-the-counter positions. Gamma exposure should be understood as an estimated market condition, not a perfectly observed fact.

Important limitations

Despite its strong results, the study does not establish a guaranteed or universally exploitable trading rule.

Transaction costs were not fully included

The primary strategy results were calculated before transaction costs. The authors note that frequent trading could materially reduce profitability. They report that S&P 500 futures retained a positive net Sharpe ratio assuming a cost of one tick, but they do not provide the same transaction-cost analysis for every individual market, including NQ.

Most return variation remained unexplained

Even the relatively strong NQ out-of-sample R² of 3.76% means that more than 96% of final-half-hour return variation was not explained by the rest-of-day return. Unexpected news, order imbalances, liquidity conditions and random variation all remain important.

The sample ended in 2020

Market structure has continued to evolve since May 2020. In particular, the growth of very short-dated and same-day-expiry options could affect modern intraday hedging flows. The paper demonstrates a historically persistent effect, but its current magnitude would need more recent testing.

Portfolio results are not individual-market results

The strongest reported Sharpe ratios were based on portfolios of several futures markets. Diversification can improve risk-adjusted performance, so the same results should not automatically be expected from a single instrument.

Conclusion

The research by Baltussen, Da, Lammers and Martens provides strong evidence of a recurring intraday momentum effect. Across more than 60 futures markets, the return from the previous close until 30 minutes before the current close positively predicted the direction of the final half-hour. The relationship was statistically significant in many markets and particularly strong in Nasdaq-100 futures.

The study also offers a plausible economic explanation. Short-gamma market makers and leveraged ETFs may need to trade in the direction of the existing move near the close, and these flows can amplify price movement temporarily before it reverses over the following days.

The most important conclusion is not that the final 30 minutes are perfectly predictable. They are not. The stronger conclusion is that closing-market behaviour is influenced by repeatable institutional mechanics, and understanding those mechanics can help traders and market observers interpret late-day price action more accurately.

Reference

All figures, statistics and quotations in this article are drawn from the following working paper. Author affiliations, as identified in the referenced version: Erasmus University Rotterdam, the University of Notre Dame, and Robeco Quantitative Investing.

Baltussen, G., Da, Z., Lammers, S., & Martens, M. (2020). Hedging demand and market intraday momentum (Working paper, 26 August 2020). Social Science Research Network. SSRN No. 3760365. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3760365

This article is educational and does not constitute financial or trading advice. Historical research results are not a promise of future performance, and the findings described are probabilistic tendencies, not reliable rules. Trading leveraged products carries a significant risk of loss.