Sounding the bell: understanding the closing auction for optimal execution in global equity markets

Dr Elliot Banks, Chief Product Officer, BMLL

First published in TabbFORUM, 7 November 2023

The closing auction is an important mechanism that plays a critical role in equity trading. The single largest liquidity event in a trading day, the closing auction has grown significantly in recent years and is a major liquidity event for any trader. Understanding the dynamics of the closing auction is therefore critical in order to achieve best execution and an optimal trading outcome.

Whilst the majority of equity markets have closing auctions, these can behave very differently, with different mechanisms, structures and liquidity profiles. In this article, we compare and contrast closing auctions across different markets, and discuss how a better understanding of auctions can help best execution, signal generation and market structure analysis.

Introduction - the growth of the closing auction

Closing auctions are a core feature of many equity exchanges, bringing together buyers and sellers without continuous matching, as well as providing a reference price for securities, indexes, and mutual funds. There has been a noticeable rise in closing auction volumes, especially in developed markets, driven by increased passive investing as well as more complex and fragmented markets. In the US, for example, the closing auction represents around 10% of trading volume. In Figures 2 and 3, we see that Japan and Europe are even higher, with 15-20% of trading volumes executed through the auction.

In contrast, for some emerging markets that figure can be significantly reduced, with markets such as Shenzhen only having 1-2% of volume trading through the closing auction.

Mechanisms

There are two common mechanisms for closing auctions in equity markets. First, the closing auction can run alongside continuous trading. Participants participate in the auction through submitting orders for the close, such as Market On Close (MOC) or Limit on Close (LOC). As part of the "closing cross process", these orders are combined, with the continuous market, a reference price is determined by maximising the volume at a clearing price and orders are executed. An indication of the volume in the auction and a potential price is provided, but actual orders are often not public. This process is in place for Canada, US and Japan.

The other mechanism, particularly popular in Europe, is where the closing auction is distinct from continuous trading. Following continuous trading, there is a call phase, where participants submit orders, but no immediate matching occurs. After the call phase, there is a price determination phase and matched orders are then executed at the reference price. For multiple European markets, the full order is visible during this time, including market and limit orders.

In both cases, participants can choose to submit either market orders or limit orders. There is therefore a trade-off between using market orders, and getting guaranteed execution at the reference price, or ensuring an execution price level with limit orders but potentially missing out on the trade. Understanding the dynamics here is clearly important for any trading strategy or algorithm.

For some markets (especially in Europe and Japan), we can understand this trade-off exactly using Level 3 data. For example, we observe that for the CBOE Europe 50 index, more than a third of the volume is executed in the closing auction with market orders, with the remainder executed through limit orders. Understanding the granular level of order matching during the auction, and how this varies over time and across stocks, can help participants better manage their choice of order type and trading behaviour.

This profile can be vastly different for different buckets of stocks (such as small cap stocks with less passive fund exposure) so it is also important to understand how these profiles vary across stocks and sectors. Also, by looking at the timing, size and price of orders submitted during the call phase of an auction, participants can also understand when price formation occurs in the auction, and adjust the timing of their strategies appropriately.

It is worth noting that not all venues provide the same level of transparency during the auction. However, even if full order book transparency is not provided during the auction, imbalance messages and indicative prices are useful information that participants can use to piece together and help answer the same questions.

Into the witching hour

Whilst auctions are important, some auctions are much more important than others. This is especially true of “triple witching” dates (when quarterly stock options, stock index options and stock index futures all expire), as well as index rebalance dates. On these dates, there is a very high demand for achieving the reference price, leading to particularly large auction volumes. For example, for the S&P 500 during the September triple witching, the closing auction volume was equal to continuous trading volume, and can be seen in the large spike in volumes in Figure 5.

Whilst it is clear that there are specific events that can lead to high closing auction volume, a natural question is what does it mean for actual executions, and how does order behaviour compare to regular days? To help answer this question, we again look at Europe. The plot below shows the notional volume of orders submitted for HSBC, both on 14th September (regular trading day) and 15th September (triple witching). Interestingly, we observe different order type behaviour as Figure 6 and and 7 shows below.

For a regular trading day, we see market orders submitted early on, and a larger number of limit orders submitted later on during the final minute of the auction. In contrast, on triple witching day a large volume of market orders are submitted very early in the auction, dominating any limit orders that are submitted later in the auction. This is the evidence that market participants have specific execution objectives during the event days, often different from the behaviour observed during the regular trading days. The insights such as this are critical for participants to adjust their trading strategy for optimal execution performance.

Conclusion: The importance of identifying varying participant behaviour across stocks and on events

The closing auction is a critical part of any trading day, with a very large amount of trading in a single event. Whilst different regions and regimes are superficially similar, closing auctions can have very different liquidity profiles, trading mechanisms and order behaviour. Using Level 3 data, users can better understand when to and how to trade during the auction, and how participant behaviour varies across stocks and on events such as Triple Witching. This helps to make better trading decisions, improve algorithmic strategies and achieve best execution.