As the rate of electronic trading in Europe catches up with that of the US, the approach of the Markets in Financial Instruments Directive (MiFID) and a stipulation for best execution, means that traditional buy-side firms and hedge funds will respond increasingly by deploying tools such as direct market access (DMA). But it will not be the only solution in the trading armoury.
Take up rates vary according to your precise definition of DMA, be it Financial Information eXchange (FIX) connectivity via a provider’s infrastructure, graphical user interface (GUI) business where a trader taps away on a terminal all day, or high frequency traders such as statistical arbitrage firms who want to participate in every tick that trades.
Greenwich Associates puts the level of DMA trading done in 2006 without algorithmic trading strategies in Europe - excluding volume sent to crossing - at 3 per cent for both pension funds and investment managers, and 6 per cent among hedge funds.
And in a survey conducted last September of 100 buy-side institutions by agency broker Liquidnet in conjunction with Greenwich, 48 per cent of European institutions “agreed” that they expected to increase their use of algorithmic trading (and by implication broader DMA), while 29 per cent were in “strong agreement” versus 50 per of UK firms being strongly in agreement.
Nicholas Wright, director, DMA and algorithmic trading at Dresdner Kleinwort in London, says: “I don’t think anyone is under any doubt about the growth of DMA business on the buy-side globally and within Europe.”
Market confusion
With the background of some 11 years at Bloomberg Tradebook, one of the first electronic communications networks (ECNs) to execute for the US markets, Mr Wright, says many buy-side firms have algorithms but have not yet fully “figured out” which algorithms to use, which orders to deploy and for which markets.
Investment banks such as Dresdner Kleinwort can differentiate their DMA offering from that offered by DMA software vendors, as the bank can offer DMA alongside research and structuring financing deals – which the vendors would find hard to match.
While Mr Wright says that nobody doubts the growth of DMA business on the buy-side within Europe, assessing some of the biggest challenges to its growth among traditional fund managers very much depends on their institutional profile, their execution objectives and how sophisticated their IT and systems in order to deploy the strategies, non-algorithmically or with “killer” algorithmic engines.
“On the DMA side there are two types of houses,” Mr Wright notes. “There are the traditional long-only funds and others. And, in the former it depends largely on the nature of the dealing desk.”
“With buy-side institutions that cover all products there can be a higher ratio of orders to trader ratio, which [generally] means they do not have the time to DMA trade in any shape or form.”
Other houses might have a slightly better ratio of dealers to orders and “attach significant value” to working orders themselves through a DMA platform. They can sometimes benefit by buying at the “bid” rather than the “offer”, thereby improving on the spread. Therefore, some shops are heavy users, while others have not “gone DMA yet”.
Tim Wildenberg, managing director and head of direct execution services, Europe, at UBS, says there are “different levels” of a DMA service. “Pure” DMA service is described as enabling a firm to trade directly on the exchange through a FIX connection. The order will go from the buy-side’s desktop straight into the exchange through a high-speed technical pipe via a member firm. But it is not algorithmic trading.
Tim Wildenberg, UBS
DMA is unlikely to replace all electronic trading methods. But interestingly for those buy-side firms that have not opted for DMA trading, there is a far better chance that such institutions will actually use algorithms, as they “remove a great deal of the manual process from DMA trading”, explains Mr Wright.
However, does using DMA – with or without algorithms – makes sense from an execution point of view given that the actual “average” transaction size of equity orders in the US equity market on the New York Stock Exchange stands at little more than 330 shares today and falling?
John Barker, managing director of Liquidnet (Europe), says: “If I were a fund manager, the last thing I would want to do is to sit there on a DMA terminal trading my own securities.”
“However, if I’m going to trade DMA, then personally I would give my trading orders over to a broker and pay perhaps two or so basis points. And, if they then make a mistake on execution, the broker has an obligation to rectify it.” Executing algorithmically and via DMA and breaking up the large orders into “child” orders might also be counter productive with traders who have the same strategies unwittingly forcing spikes in prices to their disadvantage.
Effectively it is a choice between outsourcing the buy-side’s orders to the sell-side [the investment banks], or sitting by your own desktop DMA tool to work it. But with that greater control comes greater risk. The consensus among firms canvassed by Liquidnet was that many see uptake in DMA/algorithmic trading - but not by themselves.
Trading efficiency
This constituency would “rather use a broker at the end day” who can work say 80 to 100 names of relatively small scrappy orders. That saves time to concentrate on large block orders. With the average block trade on Liquidnet being in the region of 47,000 shares between two buy-sides, it is claimed that this “dramatically” improves performance by shaving 21 basis points off each execution. That could be the difference between “being in quartile one or quartile three, if a firm was turning over their funds twice a year”, says Mr Barker.
With regard to the factors fund managers should consider when opting for DMA and algorithmic trading versus other tools, Mr Barker says: “The execution policy and the process of best execution has to be the key thing on the minds of the fund managers.”
Another obstacle to the take-up of DMA and algorithmic trading has been the capacity of order management systems (OMSs) to have a trade execution facility to DMA trade. Things are better now than they were several years ago, but the OMS applications are a slow moving beast and upgrades can take up to 12 months.
Mr Barker says that barely 50 per cent of UK firms had an OMS five years ago. Today this figure is closer to 75 per cent, while in Europe penetration is in the 60 to 65 per cent range, with Spain and Greece at the bottom. Typically a firm would only upgrade an OMS once every two years. Execution management systems, which effectively sit on top of the OMS, provide links into market venues.
Mr Wildenberg at UBS says in relation to DMA and algorithmic providers generally, the market has already started to see a “flight to quality”, which can only continue given the huge investment requirements needed to ensure the high-speed DMA pipes and low round trip latency to multiple exchanges across Europe. UBS, for example, rolls out new versions to services every four days.
Most, but not all clients, have overcome the hurdle to send orders to their sales trader through the FIX Protocol adoption, which supports an order called a new single message. While this is a simple message that, for example, says “buy 10,000 Vodafone at six or at market”, the nature of the algorithmic trading services offered by brokers requires much more information for a firm to trade algorithmically.
Alasdair Hodge, global head of e-commerce at Barclays Capital, concurs that not all DMA providers are of the same ilk and comparisons are tricky. Critically, Mr Hodge says firms must ask themselves: “Where would a [buy-side institution] want to put its trading flow through? An agency broker or a AA+ rated bank and where would they feel most comfortable?”
The hidden costs affecting smaller and medium-sized buy-side firms taking up DMA include training for the systems, functionality and general buy- side education. While Mr Wildenberg says the business of mid-sized and small firms “qualifies just as well for some of the direct execution services on offer”, the education gap needs closing to get firms “in the game”.
For hedge funds considering their choice of DMA direct execution provider, either for DMA or algorithmic trading, Mr Wildenberg adds that many of the factors that apply to institutional fund management space generally hold true - but with “subtle differences”. And, not all hedge funds are DMA users.
But it is perhaps no coincidence that the growth of DMA has followed the growth of the larger hedge funds, since incremental gains can be more difficult for them to accrue compared with smaller hedge fund players (the sub-£100m bracket). DMA can help reduce their costs.
Cost savings
Using equity finance in combination with direct execution also provides clients with much more of a combined offering, and arguably should be cheaper as one can incorporate a prime broker and/or their settlement engine.
Mr Wildenberg says: “In the case of algorithmic trading, one has to look carefully at execution quality, liquidity, access and the often overlooked point of service levels in terms of the after sales and the intra-day service.”
The buy-side should ask hard questions of its chosen DMA provider on market coverage, the algorithms offered and the algorithmic platform. While many DMA providers claim to offer good coverage, it could be that they do not necessarily have exchange memberships right across Europe. Does the provider offer Spanish DMA, South African DMA or Scandinavian DMA, for example, or does it use NeoNet as a third party in Scandinavia?
Kristian West, director at Barclays Capital, whose client base comprises many hedge funds, says buy-side firms need to carefully consider their DMA provider. Proving the robustness of the provider becomes even more critical in the wake of recent market volatility from a near 9 per cent fall in the Shanghai Composite Index.
Tim Wildenberg, UBS
