E-Trading:Algo feature – The money merry-go-round

David Adams looks at how algorithmic trading systems might be changed by volatile markets and the present downturn. Perhaps they will play a less high-profile role in the near future and more emphasis will be placed on risk, while smart order routing will move more and more to a central role?

If money makes the world go round, then increasingly that money is moving to the beat set by algorithms. One of the most striking trends in trading technology over the last five years has been the slowly tightening grip of algorithmic (algo) trading on the equities markets. The success of the algo is for good reason – they give market participants swift access to markets and enable the implementation of intelligent, automated execution management strategies and smart order routing, thus cutting transaction costs and commissions.

Algos perform a variety of useful tasks, but are most common in equities trading, where they help traders cope with the complexity created by market fragmentation, enabling orders to be split and traded at multiple venues, including multilateral trading facilities (MTFs). Crucially, algos enable traders to access ‘dark pools of liquidity’, private crossing sources of liquidity that don’t display quotes to the market but instead match orders via a technology engine. The use of dark liquidity has grown as companies have tried to minimise the impact of order flows on market risk. They also help firms comply with the latency reduction requirements brought about by the EU’s Markets in Financial Instruments Directive (MiFID) and the proliferation of execution venues it has caused.

Consequently, the use of value-weighted average price (VWAP) algos and other basic value-seeking, benchmark-defined algos has become commonplace, so much so that there isn’t much advantage in it anymore, especially if you haven’t got the fastest execution system out there. At the same time, brokers on both the buy and sell-sides have developed sophisticated algos of their own to underpin a variety of trading strategy options designed to attract clients. This combination of competitive and regulatory pressures has built up over the last five years to inspire a kind of algo arms race. The race has also caused the use of algos to spread to other asset classes such as foreign exchange (FX). But has that momentum been arrested by the market turbulence of recent months?

The dramatic events of autumn 2008 does seem to have had some negative effects. “Many of the algos that rely on mathematical models, whether volume curves, stock correlations or sector correlations, basically don’t work as well as they were supposed to, in these present market conditions” says Richard Balarkas, CEO at Instinet Europe, one of the brokers that pioneered smart order routing and is part of the group that provides Chi-X electronic execution venues. “The finesse one could achieve with algos a year ago just isn’t there now.”

Slowdown
Bob McDowall, research director at the TowerGroup consultancy, attributes a slowing in the rate of development for the more complex proprietary algos in part to the fact that there is now less capital available. “We’ve also had, over the last year, a slowdown in flows for the second and third quarter, followed by a very active last quarter, when you had the troubles of September and October 2008, because a lot of organisations were unwinding their positions,” he says. “So volumes are down. Also, the hedge fund business has suffered, particularly those that used very rapid trading techniques. So the appetite for algos is not what it was.”

Yet one might also argue that there are lots of good reasons why companies might be finding algos very attractive in the present environment. “Commission rates have fallen, people are getting less for standard equity trading and that puts a squeeze on the cost of doing business,” notes Bruce Bland, head of algorithmic research at the vendor, Fidessa. “We see algo trading being used possibly as a way to reduce costs.” He says that research he worked on late in 2008 also seems to show that when markets are volatile algos continue to achieve positive results, while the performance of human traders can become erratic. “Humans are emotional, but VWAP models are not,” he says. “They just carry on doing what they’re doing, they don’t get caught up in thinking whether or not this is the right moment to buy or sell. They don’t try to be overly clever.”

Rob Maher, a managing director at Credit Suisse and head of Advanced Execution Services (AES) sales, EMEA, also stresses the impressive performance of algos amid the turmoil of last autumn. “There were lots of questions about how algo trading would perform in a highly volatile market – would they break down?” he recalls. “The verdict, from my point of view, is that they performed admirably well.”

Maher concedes that some algo-based strategies didn’t work so well during this period, but argues that this simply indicated a need to use different types of algos more appropriate for the situation. “As volatility went through the roof and the [market risks] of spreading an order out over time rose, traders had to be more aggressive,” he explains. “We saw a movement towards liquidity-seeking, opportunist strategies. You’d see a massive shift of traders using VWAP moving to aggressive strategies that took advantage of liquidity without broadcasting their intentions to the market.” In this new environment, he suggests, different tools, like Credit Suisse’s AES algos Guerrilla and Sniper, which specialise in non-displayed trades, could offer traders what they needed. “Traders needed to be more aggressive and make a choice, rather than just spreading the order out.”

Meanwhile, MTFs, which also owe their existence and success to the growth in electronic trading, have continued to perform well. As Hirander Misra, chief operating officer at one of the most high profile, Chi-X, points out, when the exchange launched in March 2007, average trade size was €20,000. It is now €6,000. That’s a reflection of the way algos enable the use of more refined trading strategies. Orders tend to get sliced up and fed through the algo machine to avoid signalling intention to the marketplace. “That’s a trend other exchanges have seen too and it’s happening because you’ve got more and more trading going through algos,” says Misra.

“Prior to the downturn and right up to the increased volatility we had in October, with new firms and players coming in, we were seeing the pie getting bigger [even if the slices were getting smaller], and there was a lot of new liquidity coming into the marketplace,” he continues. “Then there was a period of contraction, but what we’ve seen recently is the return of the market. Turnover’s 20 per cent up and volumes are up by 42 per cent.” That’s good for Chi-X but isn’t necessarily universally replicated.

New asset classes
It does appear that the events of 2008 may have slowed algos’ march into other non-equity asset classes, especially some of the newer emerging specialisations. But TowerGroup’s Bob McDowall remains optimistic about the potential of algos away from its traditional equity base, particularly in foreign exchange. “There’s volatility there at the moment, banks are making money out of it, principally from client business, but also because they take positions to assist flow; and the FX markets are doing well,” he says. “Volumes are up and you even see FX being offered as a retail product now on a margin basis. So there are advances in that market. Investment has gone [into algos] there, to assist efficiency and at the proprietary level.”

Credit Suisse is among the companies offering algo-assisted services in the FX market. “The FX value proposition is similar to that for equities,” says Maher. “Clients are looking to trade in a more transparent and efficient manner, access multiple pools of liquidity and use execution tools to allow them to trade intelligently. In FX you have flow-based banks making risk prices, now you have an alternative transport model where you’re putting your client in control and giving them access to smart liquidity. We’ve had a lot of success on that side.”

One way algos can be used in FX is in situations where a cross-asset approach is required, such as if an equity trade needed to be settled in a different currency. “We can hedge up the FX exposure simultaneously as you’re trading the equity,” Maher explains. “Or if you have stocks trading in different markets with different currencies, you can automatically handle the FX exposures.”

Fixed income
A couple of years ago the most enthusiastic algo advocates were also sure they would start to make more of an impact in fixed income markets, but this still seems a little way off. It’s true that algos are gradually seeping into these markets, in as much as they are being used in some instances to manage intelligent pricing processes or for cross-asset trading, if fixed income products are used to hedge against positions in equities. But it is difficult to imagine algos taking the same kind of direct, primary role in what are still often very telephone-based markets as they already do in equities.

And finally there’s no escaping the fact that the economic turmoil has slowed the pace of development in algo trading systems and techniques themselves. “There’s less investment in algo trading than there was a year ago,” admits TowerGroup’s McDowall. “And because institutions are in tactical mode, vendors are in tactical mode too. Due to upcoming constraints on liquidity and regulation, one anticipates banks evolving new business models. So everyone’s in anticipatory mode. Until they see the results of more stress testing, new liquidity standards and regulations on capital, it’s going to be extremely difficult to make much strategic progress in algo trading.”

The old way of doing things is not going to be possible in a post-crunch world with new regulations and procedures. “Those external issues are more important than they were a year ago; they are driving things now,” he adds.

But as conditions become more favourable and an upturn perhaps begins next year, algos will possibly be able to escape the constraints imposed upon them by economic circumstances, and they will then surely continue their meteoric rise. They were popular for a reason and will be so again given a vibrant growing economy, although perhaps in an altered form than what has
gone before.

more features