.jpg)
While the ubiquitous discussion around the EU’s Markets in Financial Instruments Directive (MiFID) has turned the term ‘best execution’ into a buzzword for regulatory compliance, the implementation and monitoring of best execution has failed to proceed much beyond high level policy statements.
The purpose of this article is to identify the issues associated with best execution and outline some practical approaches to resolving them.
We select at random a stock (Allianz) that trades on two European markets and look at trading in that stock over a particular day to identify the problems/opportunities from a compliance/trading perspective. We then suggest a range of factors that need to be taken into account before raising a valid alert on best execution.
Notwithstanding regulatory imperatives, the implementation of best execution systems provides benefits to businesses that go far beyond compliance. Indeed although this may seem like heresy, implementation of such systems represents a good opportunity for the compliance and the trading desks to work together.
In this vein we use a widely accepted market monitoring tool (SMARTS) to provide a window into the problem/opportunity of best execution, readily accepting that most of the regulatory requirements should be able to be handled by the same tools which currently facilitate optimised trade performance, assuming they exist.
Irrespective of which department’s tool is used, such a tool must be capable of providing the ability to view and analyse the market to the tick level and below.
The tool
Before proceeding, we need to introduce the reader to a certain visualisation tool that we use and its application. Note that data portrayed in all the exhibits is sourced from Reuters.
Figure 1 shows trading in a single stock (GWT on the Australian Stock Exchange) over the course of the trading day on 26 March 2007. The x-axis represents time, and the y-axis, price. The upper edge of the green area depicts the current best bid at any point in time, and the lower edge of the red area the current best ask. The black area in between represents the bid-ask spread. Yellow areas represent a ‘crossed market’, meaning buyers are offering higher prices than sellers are asking. This is usually the case in order-driven markets outside trading hours, just prior to open auctions being conducted.
Trades are shown as circles, with the size of the trade being represented by the size of the circle. Trades are colour coded to show buyer initiated, seller initiated, crossings, and auction trades. In the actual application of this representation, it is interactive and allows many further options for the investigation of data, but for our purposes here the graphical representation will suffice.
)
Figure 1: SPREAD view of trading in GWA International on the ASX on 26 March 2007
The SPREAD view gives a useful and quick overview of the trading in any instrument over a selected time frame. In order to use this as the basis for Best Execution monitoring across several markets, we need to adapt the view from a single-market view to a multi-market view. This is what we do next.
The first tick level cross-market data views
Figure 2a mimics Figure 1 except that the stock is Allianz and the market in Frankfurt’s Xetra. Figure 2b represents the same stock at the same time but on Euronext’s Paris market.
)
Figure 2(a): Trading in Allianz on Xetra on 26 Jan 2007 (12:03pm-12:29)
)
Figure 2(b): Trading in Allianz on Paris on 26 Jan 2007 (12:03-12:29)
Figure 2c shows a view we term ‘Cross-Market SPREAD’. It is essentially an overlay of trading in a single security which is cross-listed on several markets. This display necessarily gets somewhat more complex than the single market SPREAD. The example shows trading in Allianz on 26 January on Frankfurt’s Xetra market and simultaneously on Euronext’s Paris market. A shorter time span has been chosen for the market picture here (about half an hour between 12:00 and 12:30), because the application works with shading, which does not work as well in printed form.
)
Figure 2(c): Cross-market view of trading in Allianz on Xetra and Euronext Paris on 26 Jan 2007 (12:03-12:29)
The shaded area in the cross-market view is effectively the best bid-ask on the Euronext Paris market. In general the shaded red/green areas represent a price offered in one market, but not the other. When this occurs a clear and distinct better price can be achieved in one market over the other, wherever shaded red/green exists. The black area is the combined best bid-ask spread across both markets, in this case the best bid-ask of the Xetra market.
The situation depicted in Figure 2(c) is precisely what one would expect. The primary market for Allianz is the Xetra market where the larger volume is traded. This market provides a narrower bid-ask spread, located inside the wider bid-ask spread of the secondary market. In this standard situation, a better price on both the buy and sell side would always be available on the primary, more liquid market.
What was surprising, however, was that this picture was not as common and as ‘standard’ as one might have expected.
Cross-market trade-throughs and persistent arbitrage situations
The execution of a trade outside the current best bid-ask levels is often referred to as a trade-through. On individual order-driven markets with no exotic matching rules, trade-throughs can only occur as off-market crossings. In our SPREAD visualisation, a trade-through is depicted as a circle representing the trade, located inside the red or green areas. Regular trades are always located inside or at the edge of the bid-ask spread. In operational monitoring on a single market basis, a trade-through usually immediately triggers an exception report, and is reviewed with the aim of establishing whether exceptional circumstances (e.g. volume above a minimum size) can justify the transaction occurring outside the current best available price.
A naïve view of ‘best execution’ suggests that if there is more than one available market for the execution of a trade, the market currently offering the best price should be used. If this was the case, and market participants generally operated in this way, trades would always be executed within or at the edge of the combined bid-ask spread of all markets. If a trade was to get executed outside this range, it would be a trade-through, in that it was traded outside the best available price on at least one market. We shall term this kind of trade a ‘cross-market trade-through’.
Interestingly, cross-market trade-throughs are quite a common occurrence, and they do not just occur on the secondary market, where their existence would have a much more straightforward explanation. Figure 3(a) shows an example where Allianz was bought on Xetra at USD149.68, even though sufficient volume was available to fill the order at USD149.60 per share in Paris. Figure 3(b) depicts a second example showing both buying and selling in Paris, while better prices were available at Xetra.
)
Figure 3(a): Cross-market trade-throughs on Xetra in Allianz, 26 Jan 2007
)
Figure 3(b): Cross-market trade-through on Euronext Paris in Allianz, 26 Jan 2007
Of course, if the same security is traded on several markets, certain inefficiencies between markets are to be expected, and hence certain levels of sub-optimal prices achieved by comparison with the current best available price on all markets. But the numbers do appear surprising. On Xetra on 26 January 2007, 691 out of a total of 7,714 trades were executed while a better price was available on Euronext Paris (and with sufficient volume available on Paris to execute the trade in its entirety). However, it is worth noting that the price difference is relatively low on this very liquid stock. An overall price improvement of just over EUR7,000 would have been achieved had the trades been executed at the better available prices, assuming equal execution cost, on an overall daily turnover in excess of EUR500m.
A whole range of factors can explain and in fact justify individual transactions being conducted outside the best available price on all markets, as discussed in the following section. In trading of truly the same securities on different markets, the convergence of prices between markets is in practice assured through arbitrage players, who effectively transfer liquidity between markets and in the process even out price divergences.
;)
Figure 4: Arbitrage situations between trading in Allianz across Xetra and Paris, 26 January 2007
It may therefore be more surprising still that, even in our first look at cross-market data, with example days selected at random, we find not only arbitrage situations but in fact very persistent arbitrage situations lasting for a comparatively long periods of time. In our graphical multi-market SPREAD view, an arbitrage situation is represented through a yellow area (overlapping market – as in single market view outside trading hours).
As Figure 4 shows, an arbitrage situation occurs for almost 15 minutes between 15:15 and 15:30 between Allianz trading on Euronext Paris and on Xetra. During this period, an Allianz share could have been bought for up to 20 cents less on Xetra than it could have been sold for on Euronext Paris at the same time. The available volume in the lower liquidity market (Paris) at the time is 2,000 units, so the potential profit may not be large, but it is not trivial either. This arbitrage situation disappears just before 15:30, and five minutes later another arbitrage opportunity arises.
It is worth noting that at the time of writing, we have not yet conducted larger scale studies as to the frequency and size of these arbitrage situations. However, their lifespan even in this single example appears to suggest that the use of arbitrage machines between cross-listed equities is not entirely commonplace.
Operationalising best execution monitoring
There may be many reasons for cross-market trade-throughs and arbitrage situations to exist and for sub-optimal trade prices to be acceptable to the market participants. These reasons could include differences in transaction costs on different markets, or system latencies.
They can, of course, also be the result of poorly optimised order routing and trade execution strategies, especially where individual market participants have access to more than one marketplace for execution.
In order to operationalise the monitoring of best execution on transaction levels, we need to identify and define parameters that will allow us to raise exceptions only for those cases where attention is required or desired. This holds true whether monitoring is conducted from the perspective of compliance, or whether the purpose of monitoring is the optimisation of execution performance.
The following items are variables which need to be taken into account when making the call on whether an individual ‘cross-market trade-through’ it is worth investigation or optimisation, or not:
- Liquidity constraints. Did the alternative market(s) provide sufficient liquidity at the time of trade (or order placement), to fill all or a significant proportion of the order?
- Order instructions. Did the order require an ‘all or nothing’ execution condition?
- Total opportunity cost. What is the difference between actual execution value and the value that would have been achieved elsewhere? Is it economically significant?
- Timing (latency provision). How long was the better price available for on the alternative market at the time of order placement, and how long did it last for? Was it long enough for the opportunity to have been a valid alternative?
- Market-related transaction costs. What is the difference in execution cost between the alternative marketplaces, from the perspective of the broker placing the order(s)?
- FX band. Where instruments are traded in different currencies, what are the additional execution costs associated with currency conversion?
- Historical benchmark liquidity and volatility. Where alternative markets are available with better prices, are these conditions likely to persist long enough to fill the orders placed based on historical benchmarks?
- Registry constraints. Are shares traded on different markets truly interchangeable, or are there rules related to the trading of such shares in different markets, for example manual registration processes? If the latter, this needs to be reflected through a parameter increasing the execution cost.
- How often has the situation occurred. If a particular trader is persistently executing orders at worse prices than available on other markets then action may be necessary.
By defining reasonable parameters for all the above items, we can calculate for each cross-market trade-through whether it was within an operationally necessary or justified range, and/or within order handling policies set with regard to best execution.
An additional question poses itself for the handling of client orders for regulatory compliance, in cases where internal order matching is done before orders are placed on the market. In a single market scenario, the situation is comparatively clear: crossings must be done within the current bid-ask spread, ensuring both sides of the trade achieve at least the same price available on the market. But if several credible execution alternatives exist, how is this to be addressed?
Conclusion
While it is MiFID which has turned all eyes to best execution, compliance with the directive itself is only a small part of what can be achieved by operationalising and monitoring best execution. With increased cross-listing of stocks, and increased investment in low latency trading, there are many opportunities in the practical implementation of best execution systems that enable firms to take advantage of the availability of different venues for the execution of trades whilst meeting regulatory imperatives.
Part of our purpose has been to demonstrate that one can take advantage of such opportunities by starting with an effective monitoring approach/tool. Good transaction level monitoring capabilities not only allow the identification and seizing of opportunities, but also permits immediate measuring of results from the implementation of new initiatives, as well as quick identification of changes in market conditions working in favour of or against such initiatives.
Using a well known market monitoring tool and a small randomly chosen data sample, we have first verified the effectiveness of such tools as well as the issue and the opportunity. We have also provided evidence that the scope for improvement in execution performance across trading venues appears not only to be real, but also significant enough to warrant further analysis.
Compliance with regulatory requirements imposed by MiFID becomes a by-product when taking this approach, because the standards and policies MiFID calls for are as yet not well detailed at the transaction level. Any initiative targeted at performance improvements from cross-market trading is therefore sure to exceed the steps that MiFID calls for.
