FT Mandate: Is the use of algorithmic trading strategies growing as fast as ever? If so what are the factors behind this?
Jerry Lees: We are at a turning point. For a while, there was a lot of noise about algometric models. But there was far less use of the products, particularly within traditional asset managers. But now there is a transition from talk to action. There has been a surge in the use of algometric trading tools across the board, and an increase in take up of the products on a strategic basis by buy-side firms.
I am not sure what triggered the change. In the earlier stages, a wide number of organisations had adopted the technologies, tested, integrated and talked about them. But, in practice, they hardly used them at all. Then they suddenly started to use the systems in anger.
Maybe it just took longer than expected for the organisations, people and operations to follow the theory or maybe algos just become addictive the more you use them. But, whatever triggered the change, we are seeing significantly more activity now than even a few months ago. I am told the same is true among our competitors and the technology companies.
FTM: Which products are organisations using?
JL: The buy side are using algometric models to achieve their direct market access (DMA) goals. Whether they are looking to unbundle and/or allocate a certain volume to DMA they need to be able to manage that DMA traffic in-house.
The problem is that the asset managers are not in a position to take on numerous traders to manage the increased order flow, nor do they want to. So it falls on automated order management tools with the algos to fill the gap.
Accordingly, the most popular tools by far are volume weighted average price (VWAP) and implementation shortfall, which is logical at this stage. Many buy-side firms have VWAP as a practical measure of fair implementation. A recent Edhec Advisory survey of buy side use of algometric models reported that, of those using the tools, 97 per cent used VWAP. Implementation shortfall is the other most used model - market impact is also a very important driver of buy-side trading management.
Another significant product in the Edhec survey was pairs trading. It highlighted that 28 per cent of those using algorithms worked with pairs trading strategies. This reflects the fact that, initially, algometric models were mainly used by clients with quant-based strategies - represented under the broad term “pairs” - rather than using algos to adapt more traditional long-only strategies.
As algometric trading evolves within traditional asset managers and research-based hedge funds, we will see far more emphasis on non-pairs strategies. We will also see newer products that attempt to emulate and improve on trading skills normally supplied by brokers as a part of full service broking. This will increase demands on trading desks which are already under pressure from market and regulatory trends.
FTM: What have been the most noticeable implications of these changes? How have they affected the market?
JL: We have seen a surge in the take–up of algometric trading, and that volume has had to come from somewhere. Buy-side firms are increasingly insisting that their brokers provide a range of tools and services including algorithms, DMA and programs.
The growth in volumes on the market during 2006 has not been fully reflected in traditional trading areas within brokers. DMA and algometric volumes have gone through the roof with trading flows beating the market generally. Meanwhile the traditional program desks have seen more prosaic growth. The same has happened to an even greater extent on full service trading desks.
The numbers speak for themselves - volume of business increasing heavily, size of orders diminishing as automated systems slice orders into significantly smaller bites to get closer VWAP, and so on. The result is even more dramatic growth in the number of orders executed on exchanges.
These growth statistics are reflected in the expansion of automated systems while traditional execution desks are under pressure. The automated systems generate many smaller clips and significantly lower commissions (reflecting the low use of manpower and the very low risk profile of automated trading) this puts pressure on costs and margins.
The knock on effect of all this is to put pressure on trading and settlement cost and ultimately exchanges themselves. If the exchange fee portion of costs represents less than 1 basis point of a trade and a broker is charging 15 basis points for full service ,there is a tendency to accept exchange-related fees of 5 per cent.
If the commission level for algometric trades is between 4 and 5 basis points and the number of clips inside the trade is significantly increased, two things happen. First, the cost of the trade goes up (exchange fees now tend to be transaction oriented – in cents per clip). So you begin to see algo trades costing 2 to 3 basis points in exchange fees. At the same time the commission level is significantly lower, at say 4 or 5 basis points, as I said.
Now you have a completely different dynamic. Exchange fees in this brave new world of algometric trading become nearly 60 per cent of a broker’s variable cost. It is clear now why there is so much pressure on brokers to drive down exchange costs, to look at internalisation or to look for alternative execution venues (MTFs).
When TradePoint (Virt-X) tried to persuade the market to switch flows to another venue, there was no compelling reason to do so. Now there is, with potential to reduce 60 per cent of your variable cost base by 90 per cent (as opposed to reducing 5 per cent of your cost base by 25 per cent in the case of the old TradePoint model).
Brokers must change now and exchanges must react. Exchanges will have to radically reduce their fees to address the challenges from ChiX, Posit, Turquoise, Liquidnet and other MTFs or systematic internalisers (SIs). And they will have to do it before the competitors have the opportunity to present a serious threat.
Look at the experience of LIFFE and the Bund Futures contract a few years back. Once liquidity jumps, as it did to Frankfurt, it can do so very quickly on the back of compelling technology and pricing. Once done it is almost impossible to win it back. The exchanges will be playing a very dangerous game if they ignore the threat from the likes of Turquoise this time.
FTM: How will algometric trading products and services evolve in the near future?
JL: Algometric trading is in its infancy. As the buy side starts to use these products in anger, users are becoming more demanding and to realise what they want.
Several things will happen. The models being supplied by the sell side are evolving to meet customer demand. The VWAP engines are being continually refined and new strategies are evolving.
But there are other issues. Take VWAP, for example – a very crude and, many people would argue, irrelevant or even pointless measure of performance. Most engines can produce a credible performance against VWAP – especially as 90 per cent of the performance is achieved simply by slicing the trades smaller than most human traders could handle manually. The other percentage is often largely determined by how sensible the user is in deploying the model.
For example, apply the model in a relatively orderly market on a very liquid stock and most of the VWAP algos will perform very well. They will do all the things they are supposed to do: account for the volatility curve, the momentum and market direction, apply varying degrees of aggressive or passive strategy and disguise the flow with appropriate stealth or gain strategies. But let the user apply the same VWAP in a volatile market where there is low liquidity and the performance benchmark will start to depend more on the experience and common sense of the user applying the model or the team monitoring the model’s performance back at the broker’s support desk.
So we are at the early stages in the evolution of the products available. We will see rapid developments and improvements in the quality of the products over coming months and years. In the US it is estimated that 30 per cent of flow is now handled through algorithms and that will soon be 50 per cent. This will drive the diversity and range of models available. More users will want to adapt models to suit their own needs.
There will also be a use for completely standard products that simply do the same thing (slice and manage the order in the same way). This will create a need for a range of products and tools at the user end. Buy-side organisations will not want to be dependent on a single broker solution, or even on a selection of standard products from a range of brokers. These, by now sophisticated, users will require a range of models to suit their specific needs and strategies. Moreover, these models will be consistently changing and evolving as they have to meet more demanding requirements.
For example, imagine what would happen if, over a period of time, 75 per cent of flow migrates from traders managing individual orders to multiple algometric models managing the same flow. The shape of markets and trading would be radically impacted. Every strategy we know will be impacted by the models themselves.
Now brokers, exchanges and the buy side will have to consider very carefully how they adapt to this faster-changing environment. It can’t all be done internally in any of these organisations. All of the market practitioners will have to be flexible in their approach to survive. Only the most agile will be successful.
In association with CA Cheuvreux.
CA Cheuvreux is an independent European full service Agency broker, a member of the AA rated Crédit Agricole Group.
Cheuvreux's Direct Market access product combines broad market coverage, high speed access and excellent follow through (back office and financial management). Within the group we offer access to all the major world equity markets. We also provide equity financing through CFD's to facilitate DMA clients, Algorithmic Trading and Program Trading.
Our size and distribution within Europe creates significant economies of scale which we can in turn pass on to our customers through competitive pricing.
