Into the Close: Unpacking U.S. Closing Auction Dynamics and the Impact of the Russell Reconstitution
First published by Traders Magazine, 24June, 2025
With the 2025 Russell Reconstitution just around the corner, market participants are once again preparing for one of the most intense liquidity events of the trading year. By analysing auction imbalance and level 3 data, we compare the U.S. closing auction on different exchanges, focusing on how price discovery unfolds in the final minutes of trading and the pivotal role exchange-specific features (such as D-Orders on the NYSE) have on the close. We then analyse the 2024 Russell Reconstitution to explore how auction volume and analysis metrics changed dramatically during the closing auction. For anyone trading this year's Russell Reconstitution, understanding how these auctions behave and the impact the rebalance could have is essential.
By Rob Laible, Head of Americas, and Shaurya Thakur, Quantitative Analyst, BMLL
Rebalance rush hour: inside the most liquid minutes of the trading year
The U.S. closing auction is the single largest liquidity event of the trading day. On a typical day in 2024, exchanges matched an average of $50 billion in notional volume during the closing auction, representing roughly 9% of total daily trading volume. On days with index rebalances or option expirations, the auction becomes an even more significant liquidity event, accounting for about 20% of daily volume.
One of these major rebalance dates is the Russel Reconstitution, the annual date when FTSE Russell rebalances its US equity indexes. Given the annual nature of the rebalance, it is one of the largest liquidity events in the global equity trading calendar. Last year, on the Russell Reconstitution (June 28 2024), trading volume across all Reg NMS securities was 65 percent higher than the June average, and over 220 percent higher in the final 30 minutes of trading (including the closing auction). This year looks set to be no different, with approximately $10.6 trillion in investor assets benchmarked to or invested in products based on the Russell U.S. Indexes [1].
Importantly, while closing auctions are well known as liquidity events, the behaviour of closing auctions varies across exchanges, and the nuances of how different venues absorb flow remain less well understood. For practitioners focused on execution quality, portfolio alignment, or minimizing slippage, these differences can have real impact, especially on rebalance dates.
In this article, we explore the behaviour of US equity auctions in more detail. First, we examine venue-level auction behavior across NYSE, Nasdaq, and NYSE Arca during 2025 to understand how price discovery unfolds in the final minutes of trading. We then turn to the 2024 Russell Reconstitution, using detailed auction metrics including dislocation, imbalance, and post-auction stability to assess how liquidity and auction flow shaped the closing auction. These patterns offer a forward-looking reference point for what may unfold in the 2025 Reconstitution.
Countdown to the close: understanding auction workflows
Closing auctions are used to establish the official end-of-day price by matching buy and sell interest at a single clearing price. The primary auction venues - Nasdaq, NYSE, and NYSE Arca - each conduct auctions for securities listed on their exchange. These auctions bring together auction-only orders, continuous book interest, and any other auction-eligible orders to maximize matched volume while minimizing price impact.
Nasdaq accepts several auction-specific order types in the lead-up to the Close. Participants can submit Market-on-Close (MOC) orders until the 3:55 p.m. cutoff and Limit-on-Close (LOC) orders until 3:58 p.m. Imbalance offset orders, which help offset published imbalances, can be submitted until 4 p.m.
NYSE’s closing auction includes several order types, with unique handling for each. Market-on-Close (MOC) and Limit-on-Close (LOC) orders must be submitted or canceled by 3:50 p.m. ET, unless they are offsetting the published imbalance, in which case they may still be entered after the cutoff. In addition, participants may also submit D-Orders, auction-only orders that do not interact with the continuous book and can be entered or canceled up to 3:59:50 p.m. D-Orders are not subject to the imbalance restriction and may be submitted on either side of the market.
Both Nasdaq and NYSE begin disseminating imbalance data at 3:50 p.m., with Nasdaq updating every 10 seconds until 3:55 and every second thereafter, and NYSE publishing updates every second from the start. NYSE Arca, the primary venue for ETF auctions, begins its imbalance feed earlier, updating every second from 3:00 p.m. All of this auction data can be sourced from exchange depth feeds and is made accessible through BMLL’s auction imbalance datasets, enabling detailed analysis of auction dynamics and close behavior across venues.
The last five minutes: where the auction finds Its price
A major question in understanding closing auctions is: how and when does price discovery occur in the final minutes of trading? Although the closing auction ends at 4:00 p.m. ET., understanding the final few minutes leading up to the close is critical in understanding how price discovery and formation works.
Figure 1 shows the average difference between the indicative price and the midpoint of continuous trading, relative to the final closing auction price, over the last five minutes of trading. For each plot, we choose a representative sample of 10 tickers on each exchange, with data taken from the last 2 weeks of May 2025.
On Nasdaq, we observe clear price convergence between 3:57 and 3:58. In the final minute, the indicative closing price and midpoint converge to each other, and then move in tandem toward the close, reflecting a gradual alignment of auction interest and continuous market pricing.
On NYSE, we see a similar pattern of convergence to the closing price. However, there appears to be more volatility and the final price convergence appears to be slower. This could be explained by the stocks that we picked in our sample. Another possible explanation for these differences is D-Order activity, which is a specific order type available on NYSE. We will explore this in more detail later.
NYSE Arca shows a different profile, with the indicative price fluctuating throughout the final minutes. An important point is that NYSE Arca primarily lists ETFs, rather than equities. ETFs have some important differences, and are driven by their Net Asset Value (NAV) as well as the trading price itself. This is reflected with typically less volume traded in the auction compared to equities (c. 2% of NYSE Arca ETF volume is in the auction vs 9% of NYSE equities). Note that Brokers can trade the underlying equities MOC, and then package those into the desired ETF guaranteeing the NAV risk free.
This analysis highlights a key point about US closing auctions - different stocks and exchanges can behave differently into the close. Leveraging historical imbalance feeds is critical to identifying and understanding these differences.
Held back, then deployed: how D-Orders shape NYSE’s close
To better understand the volatility observed in NYSE’s closing auction, we now analyze D-Order activity. Whilst D-Orders are not explicitly flagged, they can be inferred from level 3 order data, since these orders are placed at the worst possible price (i.e. lowest bid or highest ask) in the orderbook, but are subsequently executed in the auction. We have constructed a time profile of D-Order insertions across the final five minutes of trading. Figure 2 shows the D-Order behaviour for 3 stocks from our previous sample of NYSE listed stocks.
We see a clear surge in D-Order activity from 3:57:00 p.m. for the three stocks. Other stocks analyzed show a similar D-Order insertion profile, with most activity clustered near the end of the auction window. The sharp increase in D orders shows that participants seem to delay submitting their D-orders until close to the end of the auction period, possibly to take advantage of more imbalance information. In our dataset for 10 NYSE listed stocks (see Figure 1), 51 percent of total D-Order insert volume was observed in the last three minutes of trading.
As of August 2024, Closing D-Orders account for over 46% of NYSE Closing Auction executed volume, more than 8% higher than MOC orders [2]. Given this significant share, understanding their behavior is essential for any participant aiming to trade NYSE auctions effectively.
All eyes on the close: Russell Day’s liquidity stampede
We now turn to the Russell Reconstitution, one of the most significant scheduled liquidity events in U.S. equities, and its impact, particularly in the closing minutes. During the June 2024 reconstitution, NYSE and Nasdaq traded $275 billion and $103 billion, respectively in the final moments of trading, including the closing auction. This is driven by passive index funds, who must rebalance their portfolios at the close on reconstitution day, making the auction the preferred point in time for executing large quantities with minimal tracking error.
The chart below (Figure 3) illustrates this effect for the 2024 event, showing daily notional traded from May through July 2024. June 28, the Russell Reconstitution Day, stands out prominently. For major stocks added to the Russell 1000 (SMCI, MSTR, ALAB, PR, FIX, DELL, CRH, SN, AS, LOAR, CVNA), we observe a sharp spike in total notional traded, with auction volume swelling dramatically. In fact, on the reconstitution date, over 34 percent of total daily notional volume was executed during the closing auction, 9 percent more than that executed during continuous trading, underscoring the central role of the auction in accommodating index-driven flow, as well as trading firms unwinding positions into the close.
When the volume hits: measuring auction pressure on Russell Day
Next, we observe that auction dislocation between the final mid price and the closing auction price spikes noticeably on Russell Day across most of the analyzed Russell names, with the exception of DELL (Figure 4). The elevated dislocation underscores a temporary imbalance between liquidity demand and supply, as funds adjust their positions, causing a visible shift between the last midpoint and the actual auction closing price.
Figure 5 shows that the absolute notional imbalance is significantly elevated on Russell Day across most Russell names in our sample. This metric captures the notional value of unmatched bid orders minus unmatched ask orders at the auction price, representing the excess volume left on the book after the match.
The spike in notional imbalance suggests that the auctions on Russell Day were dominated by large directional orders and were skewed on one side of the book, likely driven by passive order flow.
Figure 6 examines auction stability by comparing each day's closing auction price to the following day's market open, measured in basis points. Despite the substantial auction dislocations and elevated imbalances observed on Russell Day, the auction prices themselves were highly stable, showing no outsized deviation at the next day’s open.
Looking ahead to the 2025 Russell Reconstitution
Our analysis highlights the distinct closing auction behaviors across different U.S. exchanges. By using level 3 data, we can see differences in price formation, order types and behaviour across Nasdaq, NYSE and NYSE Arca. This is driven by the type of instrument being traded, information available to market participants in the imbalance feed, as well as specific order types (such as D-Orders on NYSE). Understanding these differences is essential for anyone looking to trade effectively in the auction.
Turning to the 2024 Russell Reconstitution, we see how auction behaviours change significantly during index rebalance dates . We find sharp increases in auction volume, increased dislocations from the continuous mid price, and imbalances skewed to one side of the book, yet stable closing prices.
With the 2025 Russell Reconstitution approaching, fully understanding the ways different auctions behave and the impact of rebalance dates is critical for all market participants wanting to trade this event. By using information from historical level 3 market data, such as order imbalance, D-order behaviour and the timing of liquidity formation, firms can better anticipate closing price movements, effectively manage execution risk and take advantage of the surge in liquidity.