The Gateway to Algorithmic and Automated Trading

Hedge Fund Back Offices - Sinking or Swimming in the Algo Flood?

Published in Automated Trader Magazine Issue 02 July 2006

Hedge funds have been a major factor in the growth of algorithmic trading, but how are they and their prime brokers responding to the impact that algorithmic and automated trading have had on the back office?

The figures speak for themselves. More than 20% of all US equities trading was driven by algorithmic trading by the end of last year, and NeoNet has reported a spend of $230 million over 2005 on the various components needed to trade successfully. This is only likely to rise across all asset classes and regions as a result of exchange mergers, improvements in the technology available, the desire for anonymity in trading and multi asset class trading strategies.

"…our back office processing
systems are capable of
completing trade reconciliation
on all trades within
fifteen minutes of execution"

John Harrison, Amplitude's CFO

John Harrison, Amplitude Capital

The volume issue

But the rising volumes and more available usage come at a price. Adam Sussman, a senior analyst at the TABB Group, notes: "Challenges are there for firms just starting to use broker algorithms and just starting to trade electronically. They can be overwhelmed by the number of products and by knowing which algorithms are important."

However, the challenge for hedge funds goes far beyond this. It is the essence of a large part of algorithmic trading - the fact that many algorithmic trading strategies cut the order into many parts in order to get into the market under the radar of other participants and to reduce market impact - that causes the problem.

Daniel Abitbol, business development manager for Sophis' VALUE, explains: "Post-trade processing for algorithmic trading is not done the same way - there are now hundreds of thousands of executions. If all goes well, that is fine, but if there is an error the trade has to be amended and processed. Certain back office systems help, but they are still increasing the complexity of the technology. Capacity and real-time processing are the issues."

One hedge fund manager that has taken these issues very much to heart is Amplitude Capital. The firm's Amplitude Dynamic Trading Fund trades twenty instruments in time frames ranging from ultra short to short term, with all trading and back office processing completely automated. To be certain of resilience in handling the fund's extremely high transaction volumes, Amplitude depends solely upon purposebuilt technology and infrastructure it has created in house.

"Though we don't actually need to do so, our back office processing systems are capable of completing trade reconciliation on all trades within fifteen minutes of execution," says John Harrison, Amplitude's CFO. "In practice, we reconcile our positions three or four times during the day and then once at the end of the day with the fund's administrator. When you do as many trades per day as we do, the important point about reconciliation is to make sure you aren't accumulating a backlog of un-reconciled trades. Otherwise you end up with a mountain of issues to resolve at the end of the month before you can produce audited NAVs. Our objective is therefore to start each trading day with a clean slate as regards reconciliations."

Tony Freeman, executive director of strategic business development and industry relations at Omgeo

"…the main response to this has
been to raise headcount,
rather than overhaul
technology and processes"

Tony Freeman, Omgeo

Pressure

In view of auto/algo trading volumes, there is certainly increasing pressure on middle and back office systems to perform well. If hedge funds only consider the front office technology, they can easily run into problems further back and end up exposed. Although a single order becomes a large number of individual orders through the trading process, it still needs to be recognised as a single order in the back office. Funds also need a suitable communication mechanism with the broker to ensure it is dealt with appropriately.

Additionally, Robert Boardman, head of algorithmic trading in Europe at ITG, cites requests to review trading as an issue due to the sheer volume of trades: "With a quantitative manager, there is a bit more technology around the trading and broking process, and efficiency of the trade process is fundamental. Requests to review trading over a month or a quarter, with the volumes on a daily basis, are very onerous. When you have a quantitative trader, there is a lot of burden on the process."

Despite these challenges on technology, investment in the back office often takes a back seat because it does not generate revenue. For many hedge funds, this is where the relationship with a broker or prime broker gains importance. If a hedge fund is using a broker-supplied strategy, the broker will deal with most of the back-end processing. However, when large hedge funds with their own OMS want more control, this needs to be addressed. They must take into account the consequences of a much larger trade volume in their back office, both in terms of the process and timely reporting.

" We need as much aggregation
as possible so there is a quick
distillation at the end of the
trading session"

Robert Boardman, head of algorithmic trading in Europe at ITG

Robert Boardman, ITG

Some commentators feel that neither hedge funds nor their prime brokers are really addressing the volume consequences of algorithmic/automated trading. "I don't think that either group is really responding to this challenge," says Tony Freeman, executive director of strategic business development and industry relations at Omgeo. "There has been a growing divergence between the ability to execute a very large number of trades and the ability of the post trade environment to process those transactions on a scalable and risk equivalent basis. It appears that the main response to this has been to raise headcount, rather than overhaul technology and processes. I think there is also a resultant suspicion on the part of regulators that this situation means that as trade volumes are rising, so are the risks."

The problem has much to do with established practice and attitudes. While conventional asset managers and custodians might tolerate manual back office processing of perhaps 5% of their total trade volume, for the alternative asset management industry the figure is often much higher. Manual processing of 30-40% of trade volume may be regarded as acceptable. In part that is due to the greater complexity of the instruments traded by the industry, but it still represents a huge number of trades that may develop into a substantial backlog of unsettled transactions.


Regulators have already openly expressed concern about this as regards certain asset classes, such as credit derivatives. However, what practical remedies they may have as regards setting a maximum percentage of unsettled trades is another matter. While they can obviously influence the regulated broker dealer side of the fence directly, this only exerts indirect pressure on the hedge funds, who may simply select other trading partners in response.


Alternative and multiple asset classes

Coupled with the issue of increased volumes is the rise in alternative asset classes being covered by algorithmic and electronic trading, and also the different types of algorithm available. There has been a rise in the last 18 months in the number of asset classes being traded algorithmically. In Europe particularly, there has been significant penetration into the derivatives market. While exchange-traded derivatives have the same structure as equities, thus requiring little variation in processing, OTC presents more challenges. FX and bond markets suffer from being quite opaque, and the markets need to reach an increased level of transparency for algorithmic traders to take full advantage.

Multiplication of products also means that there is a need to get definition of those products. It may be less relevant for futures, for example, but with hundreds of strikes and maturities, a database is needed in order to trade options properly. Data providers such as Reuters and Bloomberg may have a role to play in overcoming this issue, according to Dr Giles Nelson, director, algorithmic trading at Progress Software. He adds: "Hedge funds can use some of the same principles used in equities for pricing and market-making. The business use case and high level processes are different, but the technology can be the same

Daniel Abitbol, business development manager for Sophis' VALUE

Post-trade processing for algorithmic
trading is not done the same way -
there are now hundreds
of thousands of executions"

Daniel Abitbol, Sophis

In addition, swaps bring a separate set of challenges. ITG's Boardman notes: "The transaction process is somewhat more complicated - we need to focus on the back office. We need as much aggregation as possible so there is a quick distillation at the end of the trading session. It is a somewhat manual process, and it is very uncomfortable for everyone involved. There are moves in the industry to create an automated platform for swaps give-ups - ICAP is involved in this. The industry has needed this for a long time."

Algorithmic trading across multiple asset classes is now a hot discussion topic. However, while the trading desks may be raring to go, the back office is still well off the pace. One of the main bottlenecks here is the order management system, which typically acts as the focal point between front and back offices. Many individual order management systems - particularly those in house - are already creaking under the strain of algorithmic/automated trading. The huge transaction counts now prevailing are way beyond their original design specification, particularly in the case of participants active in US equities.

However, a further issue is that many order management systems have been built to be asset-specific. In multi-asset algo/auto trading this requires coupling and synchronising multiple order management systems to feed the back office. With some individual systems already running well beyond their design capacity limits, the multi-asset algo/auto prospects for the back office don't look particularly bright.


"…hedge funds still need
a back office system
that can cope"

Dr Giles Nelson, director, algorithmic trading at Progress Software

Giles Nelson, Progress Software

Risk and cost

With a relatively new strategy such as algorithmic trading, one might expect that there would be new risks to be overcome. One of the principal issues is that of risk management for the middle office. One consequence of executing a vastly increased number of orders in a rapid fashion over the course of the day has been that positions can change more radically than previously. The standard is for risk management systems to deliver an end of day report, but this is not really acceptable at this level - funds need the ability to look at risk and exposure intraday, preferably in real-time as trading takes place. Then, if a fund goes beyond a particular threshold, it will have the option to reduce its exposure. Some firms are using some of the algorithmic system technology to do this automatically.

In his February 2006 report on hedge funds' quest for alpha, TABB Group's Sussman asks: "It seems that the use of algorithms is resulting in pressures on the IT infrastructure. Taken together, does the cost of upgrading infrastructure, OMS and trading systems, processing trades and then analysing them actually make sense?"

Progress's Nelson believes that it is all down to the usage of particular firms: "Algorithmic trading is not for everyone. There are small hedge funds that are reliant on brokers, which may have been formed by three or four high net worth individuals looking to get very good returns. At the other end are more traditional asset managers and unit trusts that use it for productivity, not money making. Each organisation needs to look at the consequences for their systems, but hedge funds still need a back office system that can cope. Algorithmic trading is worth it as long as due diligence is done and hedge funds recognise why they are taking it on."


Future imperfect?

So what does the future hold as algorithmic trading becomes more entrenched in the strategies of hedge funds and volumes continue to increase exponentially? Some form of further investment in new order management systems seems inevitable if a processing log jam is to be avoided. With regulators already taking a keen interest in the alternative asset management industry, this is much to be desired.

Another development that could ease matters considerably - especially in alternative and multi-asset algo/auto trading - is new messaging standards for the back office. "Much of the messaging protocol focus we have seen so far relates to executing trades," says Omgeo's Freeman. "Far less effort has gone into post trade messaging standards. This is already improving due to the efforts of bodies such as FPL and FpML, but they still have some ground to make up - especially when you consider the pace at which the alternative asset management industry innovates as regards tradables."

All told, with change being driven by technology and regulation, profit and efficiency, the back office - be it proprietary or provided by brokers - will continue to have its work cut out to keep up for some time to come.