Jon Carp, Head of Alternative Execution Sales Europe at CA Cheuvreux
This raises some interesting questions relating as to how execution algorithms and automated trading systems react to such sharp market and price movements. Surely, when algorithms are mathematically formulated, it is almost impossible to build in the prediction analysis required to cope with the uncertainty we have been experiencing recently. Algorithms use history as a guide to their future behaviour so, when markets are acting in a way perhaps never seen before, how do they cope?
Tuning the engine
I believe it is widely accepted that 'black boxes' have exaggerated the market movements over the past year due to this new market behaviour and a lack of comparable historical market data. The wide spread of percentage gains and losses recorded by European-based quantitative hedge funds in 2007 perhaps prove that machines need fine tuning.
While the majority report gains for the year, far more show
negatives for the final six months when the volatility really
So if algorithms are experiencing unknown territory it would follow that their usage would reduce. Why would traders trust algorithms when they are operating in a potential minefield?
Interestingly, our experience at Cheuvreux is the opposite. The steady rise in usage of our algorithmic strategies, by our clients, over the first half of 2007 was suddenly accelerated during the second half, and the chart sharpens even further as we enter 2008, with huge daily market movements. Perhaps the previous 12 months has provided more relevant data for the algorithms to perform better. Maybe traders prefer to put 'noise' trades in the machine while they focus on the others. Evidence from our own clients suggests the former. It is noticeable that the average size of the 'parent' order has substantially increased.
Breadth and depth
While it is true to say that we have invested in making the
algorithms available on more EMS and OMS providers, we have also
delivered more strategies that are benefiting from previous
trading data. Likewise, we offer algorithms on all markets we
cover in Europe and the US, and indeed a wide universe of stocks.
While others may offer algos on blue chips, we look to
distinguish ourselves by not only offering these but also algos
on mid and smallcap stocks. For these, starting with history then
analysing intraday and real-time data as early as possible is
imperative due to the stocks' varying liquidity.
The increase in usage of our algorithmic trading products can also be put down to other factors. We continue to invest in people who have deep and wide experience of algorithmic trading and act in a consulting role with clients. This benefits all concerned. By consulting with our clients, we're better able to understand which strategies are best suited for what they are trying to achieve. In many cases we have found that clients change their initial favoured strategies after experiencing different algorithms.
Service begets feedback
Although providing a consulting service incurs cost, we
appreciate that, together with performance, customer service and
confidence is a key factor in client loyalty. From our standpoint
we gain valuable client feedback, which enables us to continually
fine tune and improve the algorithms. The challenge does not end
here, however. European markets should be readying themselves for
further upheaval. Although we can not accurately predict future
market volatility, we can plan for change in market structures.
MiFID came, ultimately with no more than a murmur. Not much
change in trading apart from Chi-X offering a choice of venue
and, therefore, the arrival of smart-order routers (SORs) to
Europe. But we are aware of other venues, be they exchanges, dark
pools or a mix of the two, being launched over the coming months.
The story is only really starting. The immediate future requires a combination of technologies that allow algorithms to work in conjunction with the SOR over a variety of liquidity pools, 'pinging' for and/or placing for liquidity, while constantly watching all venues to ensure optimum performance. Fortunately, we have experience of this in US markets where the proliferation of venues has been dramatic, especially since Reg NMS. Therefore, the challenges for Europe - and technologies deployed here - are not necessarily new, but the ideology is.
As I have written in previous articles, the challenges for providers of algorithmic strategies do not simply exist in producing products equal to competitors. A large investment is required to produce them in the first instance and then to have a continuing enhancement programme that keeps performance in line and above expectations of clients. However, our experience shows that to exceed expected returns requires more. Producing distinctive and proprietary algorithms that clients actually want, benefit from and utilise latest technologies is one aspect, but add this together with excellent customer service and you approach the correct path forward.