From the May 01, 2007 issue of Futures Magazine • Subscribe!

Automated trading: An exchange’s best friend

Peter Green has been there from the start. The company he runs, Kyte Group Ltd., opened its first trading arcade in 1997 with roughly 30 former floor traders from the London International Financial Futures and Options Exchange (Liffe). “We opened it across the street from Liffe so that traders could move between the floor and our dealing room very quickly,” he recalls. “At the time, most traders worked on the screens during quiet periods in the market, but almost invariably they would run — literally — back to the pits if there was any real excitement in the markets.”

Those days are long gone, but Kyte continues to run arcades across London. Dublin-based Geneva Trading runs similar operations in Chicago, with traders acting as commissioned employees, splitting profits with the house.

Such operations have surely contributed to the growth in volumes the industry has enjoyed through the past few years, as have increasingly sophisticated directional trading houses that use sophisticated algorithms to identify trends and reversals.

But these days, when we talk about algorithmic trading, we usually mean teams of mathematicians, physicists, and computer geeks looking to capitalize on very tiny aberrations in and among markets. This new breed is responsible for the tight bid-ask spreads that we on the retail side now enjoy, and is beginning only now to venture into directional trading.

Although no one has managed to quantify algorithmic trading in absolute terms, everyone agrees it’s the driver behind the surging volume on exchanges worldwide.

“Algo trading is now more than half our traders’ business,” says Green. “We need to nurture this and look for the next wave of trading innovation.”

But his competitors are increasingly deep-pocketed banks like Netherlands-based Fortis, which owns old-line Chicago clearing firm O’Connor & Company. Under the O’Connor banner, Fortis runs a massive trading room across the street from the Chicago Board of Trade (CBOT).

The facility rents space to a handful of former floor traders and groups of algorithmic traders who tend to work in teams, something Green says has been the modus operandi of choice in London for years.

“There is a new breed of trading firm that is engaged in automated, worldwide and multi-asset-class trading, and these firms are becoming incredibly successful,” says Linda Bracken, managing director of YJT Solutions. “These shops might only have 20 people, but they want top-of-the-line solutions and maximum trade speed. Many people would be stunned if they knew what these technology budgets are.”

The big winner so far appears to be the exchanges themselves, which aren’t so much competing for algorithmic business as they are harvesting it to keep their liquidity pools deep, liquid and attractive to all players. Futures commission merchants (FCMs) and technology providers have no doubt benefited too, but on that end, the competition is more fierce.


The only place algo trading hasn’t taken hold yet is commodities, which are still freshly electronic. Grains, for example, only went electronic during regular trading hours last August, and options won’t trade electronically during open outcry hours until later this quarter.

Suzanne Dence of the IBM Institute for Business Value points out that the securities involved in grain trading, and commodities in general, is more complex than in financials.

The CBOT, understandably, is doing its bit to help the algo boys up the learning curve with a series of training courses designed just for them.

“We go through fundamental factors,” says David Lehman, managing director and chief economist at the CBOT. “We cover the basics: what are inter-commodity and intra-commodity (calendar) spreads, how do they work, how is the corn crush different from the soybean crush, how are grain calendar spreads similar to the yield curve, what government reports impact grains and how do you access them and what is the delivery process? Things like that.”


Some exchanges, like Eurex and the Chicago Mercantile Exchange (CME), offer “co-location” sites where algorithmic traders can physically place their computers in an exchange data center for nanosecond execution. YJT Solutions also has been building data centers next to exchanges for years and offers 24/7 networking support. The CME also offers member rates to European prop shops that meet certain criteria, and it recently agreed to alter its code to protect anonymity, but all exchanges are leery of offering too much attention to any single user group for fear of alienating others.

The CBOT, for example, doesn’t offer colocation sites, but does meet regularly with a number of prop shops who comprise a so-called “professional trader” user group committee, and also runs an electronic market-maker program for specific new and existing products. These specific programs generally involve fee waivers and other enhancements in return for contractually agreed-upon minimum levels of bid, ask and size.

“Creating liquidity is a tricky business,” says Bob Ray, CBOT senior vice president of business development. “Not least because there’s a question of where you cross the line from being a liquidity maker to a liquidity taker.” Market makers, he points out, face a dizzying array of risks and opportunities, with typical market-makers active in scores of correlated markets at any time.

It’s common, for example, for a trader to make markets and hedge risk in a variety of interest rate products and to hedge them all on a portfolio basis. It’s also common for such activity to make it more profitable for market makers to give up a tick on the leg in the market they’re obligated to make to both execute a more profitable strategy and meet their market-making obligations.

“What it boils down to is that instead of waiting for the client to come in and hit them on the bid, they’re actually going in and lifting the client’s offer,” Ray says. “If that starts happening, we might want to ask if maybe that particular market has matured to the point where we don’t require as extensive of a market-maker program as we may need in a less mature market.”


Exchanges may be unwilling to offer too much special treatment to algo shops, but FCMs and technology providers have been ratcheting up their courtship of those businesses.

“That’s the biggest change over the past two to three years,” says a salesperson at one of the largest market participants. “It used to be you had just a handful of FCMs — like the old Refco, or Fortis, or Man, who would go for prop shops as customers, while most didn’t get involved because these shops can disappear overnight. But now that the Goldmans and Deutsches see how big this business is, they’re all looking at that space.”

So far, they’ve managed to avoid a price-slashing war, and are instead competing by offering massive bandwidth and all the bells and whistles they can muster, usually in cooperation with one or more of the array of technology providers that has evolved along with the industry. YJT’s sales quadrupled last year, largely because they have been able to establish themselves as the Team Ferrari of network support and connectivity.

“There used to be just one or two ways to connect to an exchange and set up a network,” says Bracken. “Now there are multiple ways to get at these markets, and the complexity of network maintenance is growing as well. Everybody wants to shave milliseconds and increase network reliability.”


Many of the bells and whistles being offered now to keep algo traders happy were themselves cutting-edge algorithms just a few years ago – a fact that says much about how quickly the space is evolving.

“There are basically two types of algorithms,” Dence says. “One type creates efficiency in the markets and the other type uses arbitrage to generate alpha.”

She says the use of algorithms to generate alpha is leveling off in mature markets like currencies and equities, but will continue to grow in fixed income and derivatives before coming of age in commodities.

“It’s pretty easy to see where the potential for growth exists and where it doesn’t,” she says. “If you look at the maturity curve of any product, you see that the price clients are willing to pay correlates to value, and the majority of the algorithms used in equities are now given away for free.”

Plus, she adds, the quest for alpha is a zero-sum game. “You have too many people using the same algorithms,” she says. “The model is broken — at least in equities.”

In theory, that’s good news for directional traders because it means correlation among markets is on autopilot: the professionals have built machines that will continue to snap any markets that get out of line back into place, and the rest of us can get in and out at will.

There is, of course, a dark side: some algorithmic traders have developed products that game the current algorithms by placing orders targeted to distort electronic order books and cause a brief but violent move in one direction or the other, with the intent of making money on the correction.

Such activity, however, is nothing new. It represents an electronic version of the old floor-traders’ game of “running the stops,” and retail traders simply need to be aware of the landscape.

“The emerging frontier that has not solidified in terms of algo maturity is commodities,” Dence says. “Algo trading in commodities will be a longer wave because the securities themselves are more complex, but [OTC] fixed income isn’t really that difficult at all, and broker dealers have made statements to us that within five years time they should be able to handle virtually every instrument that exists.”

Technologically, yes — but what about politically? Electronic trading of OTC bonds has only taken hold in Europe because the MTS platform was able to break the cabal of “usual suspects” led by Goldman Sachs and others who keep the markets on the phones, where margins are fat and juicy.

London-based Datamonitor has reached similar conclusions. In a study released last December, they pointed out that most new algorithms are focused on analyzing transaction costs, and predicted that the new wave of arbitrage-oriented algos “will follow a different evolutionary cycle to that of equities (and) target greater intelligence to combat competition.”

Within the financial sector, they project the take-up of algorithmic trading will peak first in options, then futures, then currencies and finally fixed income, which they say will be “last to fully move because of the dealer set up in this space.”

Dence doesn’t deny the incestuous hold of major broker-dealers on today’s market, but says the drive for best execution and growing competition from exchanges will ease the grip over the coming decade.

“We believe the exchanges will expand their capabilities now that they are for-profit,” she says. “Over time, they will be able to scoop up these OTC trades and clear them.” Her prognosis: “70% of all OTC bond trading will be done electronically by 2015.”

And, we don’t need to add, electronic trading generally leads to algorithmic trading.

“For now,” she adds, “the big growth in algo trading will come on the efficiency front because both the United States and the European Union are mandating best execution, and algorithms are more capable than humans at achieving best execution in many cases.”

She even says the best-execution provisions will find their way to futures, with contracts mandated to be fungible across all platforms, and brokers mandated to get the best possible price for their customers. “That’s where the laser of maturity points,” she says.

Both IMB and Datamonitor see algorithms taking greater hold among directional traders, who are already beginning to incorporate news feeds into their automated strategies.


The question in Chicago is whether grain traders can harvest their knowledge in the new arenas. “Here in London, the floor traders who made the transition were usually spreaders and others who tended to stay market neutral,” Green says. “If they adopted a market-making approach and operated on high volume/low risk they tended to be more successful,” he says. He estimates a trader failure rate at 50% the first two years, but says today’s prop shops have developed sophisticated trader development schemes to ease the transition, or, more coldly, cull the herd of weak members. “Since 2000, our growth has been consistent and steady,” he says. “Trader failure rates have fallen dramatically and we now have a fairly stable community of over 200 with the kind of turnover you would expect in any firm of that size.”

comments powered by Disqus