FT Mandate: What are the main drivers behind the increasing demand for algorithms from the buy-side?
Tony Bayliss: The obvious drivers are costs, and the efficiency of algorithms. Algorithms were originally created to assist sell-side dealers in coping with large volumes. They are a great way to deal with that – you divide up your order book, giving more time to focus on where the trader can add more value.
We are also seeing a lot of usage where you have very tight constraints on the securities you are trying to trade. Obviously, an algorithm will be watching all the time and will grab as soon as a bid appears, whereas a human trader may be off the desk for a period of time.
The time to market is a lot quicker for algorithms. It is quicker to enter algorithm parameters into the order system. Also, there is a feeling that there is greater anonymity by using an algorithm. You are not dealing with a sales trader in the middle of the trading floor, who then shouts it across to a trader. So sometimes people will have more confidence putting in price sensitive information.
FTM: Are traditional asset managers catching up with hedge funds in the uptake of algorithm trading?
TB: We are starting to see that now. A lot of firms have a target of putting 10-30 per cent order flow through algorithmic trading. As a general rule, trading styles are less aggressive for asset managers than for some of the hedge funds but the general rationale behind using algorithms is often the same.
FTM: The need for gaining access to the growing number of dark pools of liquidity has resulted in the development of new liquidity-seeking algorithms. How do they work?
TB: Smart order routers work by having an understanding of each of the liquidity pools and how they work and interact with flow. Each venue differentiates itself by having its own methodology for interacting with the actual flow that is there. Negotiation takes place in different ways. A smart order router needs to understand each and have access to all of them.
Depending on the aggressiveness of your liquidity-seeking algorithm, it may seek only to interact with dark liquidity –sending in hidden order types to see if they are matched on the other side. To get more aggressive, you can interact with order book flow or you can post liquidity on certain venues.
FTM: And what are the challenges for sell-side firms trying to differentiate themselves in this area?
TB: It really comes down to speed, which is crucial; knowledge of where things are trading; and building up that knowledge over time.
Because speed is important, you want to target venues where you have seen liquidity in the past or where liquidity exists at the moment. Because there are so many venues you can only ping so many at a time before you have to prioritise which will target. Pinging is sending the minimum size order for that venue and seeing if you get a fill. The actual knowledge side is also crucial. At Citigroup we own Lava, which has the concept of a dark book. This effectively allows clients to enter the full order size even if they only send a small part to market. Lava has the knowledge that a large order exists so it can match large buyers and sellers of large blocks. It gives you a higher probability of finding the other side for your whole order size.
FTM:Recently there have been significant developments in the use of algorithms in foreign exchange. Where is the demand for these products coming from?
TB: All international fund managers have to manage their FX positions so there is quite significant demand. The FX market is significantly different from the equities market in that it is all over-the-counter, so effectively you have major counterparties simply posting up their own spreads, usually real time, that you can interact with, and those will widen depending on size of liquidity they are trying to access. It is more akin to risk trading, where you go with the best price and generally can do the whole order at one time.
FTM: How do you see the future of algorithm trading in Europe after the arrival of MiFID at the end of the year?
TB: Europe will see the market fragmentation that they already have in the US. There will be multiple trading venues to secure best execution. Already, with Project Turquoise, a consortium of investment banks including Citigroup will be setting up a competing pan-European exchange.
It will be necessary to build up an international best bid offer. If you look up, say, Vodafone on the FTSE at the moment, you will have bids and offers. A best bid offer can recreate that from all the venues, and that is something any algorithm will have to monitor. Various data providers are looking to do that already.
There is also an expected increase in dark liquidity. There are many more competing venues, and sell side investment banks are opening up their internal crossing networks to allow the buy side to interact with their own internal orders.
It is also worth mentioning that algorithmic trading is now diversifying into other asset classes – FX, listed options, fixed income, and a multi asset algorithm at the portfolio level.
In association with Citi.
Citi is the most complete financial partner to corporations, financial institutions, institutional investors and governments in the world. As a global leader in banking, capital markets, and transaction services, with a longstanding presence in more than 100 countries, Citi enables clients to achieve their strategic financial objectives by providing them with cutting-edge ideas, best-in-class products and solutions, and unparalleled access to capital and liquidity. Citi also provides a broad range of financial products and services to individuals through brand names including Citi Smith Barney, Citibank, and Banamex. Additional information may be found at www.citigroup.com or www.citi.com.
