Much has been written in this and other publications about the ever-growing availability of trading algorithms available from sellside firms and the impact of MiFID. Less, however, has been written about the implications for the sellside in general and the knock-on effect on the buyside.
The sellside can be broadly split into two categories:
1. Those that have a wide selection of market membership and offer a range of services, including electronic (both DMA and Algorithms) trading to their clients and…
2. Those that offer niche trading and services to their clients and are either members of one, a few, or no markets.
Sellside category one
For the first category much has to be considered, as the client base of both buy and sellside must be satisfied. Let us consider the 'Best Execution' directive from a pure DMA perspective.
Until MiFID is implemented in November, the client simply sends an order using symbology to denote the exchange upon which the order should be placed. This is, in most cases, where the liquidity for that stock is either complete or greatest. However, post-MiFID that liquidity could be split between various exchanges and multilateral trading facilities (MTFs). The DMA provider will need to provide access to these various liquidity pools, including the new venues…. but that alone is insufficient. The DMA provider must offer its clients (as now) the option of being able to denote their preferred trading venue for any single order. However, it must in addition offer them the option to trade 'market order' from a consolidated quote, provided by information providers such as Reuters and Bloomberg, across venues. Here, the order may have to be split between venues depending on the price and size quoted on each. To achieve this, investment is required in very efficient Smart Routing tools that will not only do the job but also integrate into existing and future systems. Fortunately for Europe, we can learn from the US experiences of fragmentation and subsequent technology developments.
We should consider the impact of this coupled with the fact that European exchanges currently have varying trading and settlement costs. For example, this means that a market order that is split by the order router in a simple 50/50 ratio will have differing trading costs and two differing settlement costs - the non-existence of a central clearing facility like the US DTC facility means that the DMA provider will potentially see a large increase in settlements and therefore costs.
And what of Limit orders? Where does the Smart Router place these if a venue has not been specified? Logic tells us that the router must be able to calculate, historically, the venue that is, and has been, the most liquid for this stock, to maximise the fill probability.
Finally, if the DMA provider has order flow from both retail and institutional clients, how should it maximise best execution for both categories of clients? Perhaps by building a crossing engine, with the smart router deciding if this is an overall beneficial execution facility for a singular order. Bear in mind that if the broker can cross an order at the best available prevailing price, then they reduce trading and settlement costs and ultimately client commissions.
"BUT they have to accept they are suddenly paying a brokerage commission..."
Sellside category two
So what is the impact on the second sellside category? For example, consider a broker who specialises in UK stocks only, currently receives orders by phone and does a good job of executing through their own UK market gateways. A quick look at the London Stock Exchange membership book shows that many such brokers exist. What if post MiFID they are given an order to buy 500,000 BP over the day and if the real liquidity of BP is split between, say, London, Euronext, Xetra and new venues such as Turquoise and/or CHI-X? Currently they only have membership of London, so do they take on memberships of some or all European exchanges/MTFs where liquidity may exist in the future, thus incurring monetary, systems and personnel investment? Or do they connect to a 'house' in the first sellside category, where the impact analysis and investment is done for them, BUT they have to accept that they are suddenly paying a brokerage commission on trade execution, whereas previously they were paying only exchange fees and settlement charges?
These sellside firms will fall under Receiving and Transmitting of Orders (RTO) rules so cannot totally relinquish their MiFID obligations. They must use entities whose execution arrangements will enable the RTO to comply with its own obligations and must also monitor the execution quality of those third party entities on a regular basis. Both categories need to have an execution policy, but the key point is that those in the second sellside category must always be able to justify their broker selection policy. So for the sellside there are still outstanding decisions to be made regarding the necessary levels of investment and the services to be offered.
How does all this impact the buyside? Firstly, and perhaps most obviously, broker selection criteria will be paramount. Do they go straight to brokers who have a wide membership and connectivity base - hence missing out the 'middle man' who has previously serviced them well with expert, albeit manual, execution but without the financial resources to develop wider ranging systematic offerings. If they continue to use such brokers how do they ensure, for the benefit of their investors/clients, that the chosen broker has the means to monitor and prove execution quality.
Underlying all of this, sellside and third parties continue to develop new execution algorithms to attract and retain buyside clients. In this limited space will not opine as to which algorithm will be the most successful post MiFID but I will venture to suggest that a number of simplistic factors will be predominant in the buyside's decision making process. For example, can the algorithms 'search and play' on all the available pools of liquidity? Currently, for a basic VWAP strategy algorithms rely on a historical volume curve from a single liquidity (or trade reporting) source…. in the future and, with multiple pools, will these curves be an accurate reflection of trading patterns? Does the broker have direct membership of the pools or are they routed through another member? If the latter, what is the latency impact? This raises the question of how efficient the algorithms and internal systems are in dealing with differing latencies? With the supply of market data from the various execution venues making up the prevailing consolidated quote from which algorithms will work, how accurate will that quote be? Maybe the first question for the client should be 'Do the broker's own traders use and rely on the algorithms being offered to clients?' - answers may vary but flexibility could be the key.
"What is the latency impact of
Without doubt, much remains to be done in all areas. More substantial companies are probably relatively advanced as regards their post-MiFID offerings - many key decisions will have been made and others continue to evolve. Investment to meet current and future requirements is recognised as essential and is in place. By contrast, smaller specialist sellside firms are at varying stages of analysing what services they will need to offer and indeed how they will offer these within the required regulatory framework. Buyside firms have some key decisions to make in terms of broker selection, the trading services they plan to use and who offers them - all of such decisions have to be justified on an ongoing basis.
As I said at the beginning of this article, much has been written - but we are now getting to the point where much must be delivered.