What role are algorithms playing in the continuing trend towards exchange-traded funds (ETFs)?
At Citigroup, many of our clients use algorithms to assist in the execution of ETFs. However, these transactions have focused primarily on the major liquid ETFs in the US market. This is consistent with what we advise our clients on the best way to employ our algorithmic suite of products.
If the trade is expected to be a significant proportion of available volume, or if the ETF is less liquid with a wider bid-ask spread, the optimal trading route is still likely to be dealing directly with the market maker. An ETF market maker can create and redeem the ETF and provide the required liquidity.
Dealing in an open manner with a trusted market maker, giving them sufficient time to rehedge their underlying position, is likely to have less impact and enable them to source sufficient liquidity. This additional liquidity above normal trading volumes is unlikely to be made available directly on the exchange.
To what extent are asset managers and institutions basing investment strategies on algorithm-based quantitative research vehicles?
The high-frequency trading space has been dominated by niche players and hedge funds. For the more traditional asset managers and institutions, the level of uptake has so far been limited. But we are starting to see signs of increased interest.
New funds are being set up, and discussions about the required infrastructure are taking place, to enable institutions to start exploiting short-term market movements.
What are the obstacles to further take-up of algorithmic trading by European buy-side companies (for example, greater connectivity, FIX interfaces or client education)?
Many buy-side institutions still lack the technological infrastructure necessary to benefit fully from algorithmic offerings. More than 70 per cent of equity orders are sent to Citigroup via FIX. But that still leaves a significant proportion of clients for whom sending orders and receiving executions is still a partly manual process. Integration of all the different offerings within the order management systems (OMS) is also a significant challenge. For most vendors, it will take some time.
Institutions with smaller dealing desks, and even those where trading is carried out by the portfolio managers, often do not have the time to load and then monitor their own algorithmic transactions. For these institutions it is still far easier to pass on that task to a broker.
Transferring a trade to a broker, rather than an algorithm, also enables the institution to pass on some of the responsibility for the quality of execution. This gives the buy-side the confidence that there is an additional pair of eyes checking the style of execution, monitoring performance and sourcing liquidity via traditional sales trading.
What are the factors in choosing an algorithmic trading strategy?
Some factors can be calculated based on where and how the security has traded historically, and some that are based on the views of the fund manager or trader who has made the investment decision.
Bid-ask spread, availability of liquidity, volatility, depth of order book and momentum can all be modelled and used to narrow the choice of suitable algorithms. Views on alpha, investment time horizon and how the stock will react to news or market events are all more subjective but can have a dramatic impact on how a security will be traded.
The performance of each algorithmic execution should be recorded after each transaction and the style or reason for investment should be noted. This should then form part of an iterative process so that, for a given security and investment style, you can achieve the best price.
The primary performance goal, when executing a trade, may not be simply to buy low and sell high. There are other constraints on dealing desks and fund managers that may tightly limit their choice of algorithm. For example, a fund may be benchmarked to the close price – so, to minimise the risk of missing this benchmark, a Market on Close (MOC) algorithm may be an obvious choice.
What challenges do hedge funds present to the providers of algorithmic trading systems?
Hedge funds tend to have a much shorter time horizon on their investments. They often react to news alerts or capture short-term price movements with higher frequency trading models. This places a stronger emphasis on more aggressive, short-term trading models where the time between decision and execution becomes all the more important.
As more institutional assets flow into hedge funds, will they become the primary buy-side users of algorithmic trading strategies? If so, why?
In the US we have already witnessed a significant narrowing of the gap between hedge fund and institutional algorithmic trading flow. US trends tend to spread across the Atlantic and we anticipate a continuation of the rise in hedge fund algorithmic activity in Europe. The higher turnover and higher frequency trading strategies of hedge funds is likely to accelerate that trend. We expect to see hedge funds becoming the primary users of algorithmic trading strategies within the next two years.
In association with Citigroup Corporate and Investment Banking.
Our Corporate and Investment Banking group achieves the extraordinary for our clients around the world. No financial institution is more committed to understanding and advancing the goals of its clients – our diverse and talented staff in some 100 countries advises companies, governments, and institutional investors on the best way to realise their strategic objectives. We create solutions for and provide the broadest possible capital and market access to thousands of issuer and investor clients, and no institution better executes the increasingly complex payment and cash management solutions required in today's global economy. Additional information can be found at www.citigroupcib.com.
