Discussion in 'Automated Trading' started by ASusilovic, Aug 31, 2008.

  1. Trading Wizards

    Arcades rack up high volumes, spend more on technology and attract IT talent who use "gray box" strategies.

    Few of the restaurant diners and pub-goers on Islington's fashionable Upper Street in London have heard of Kyte Group. But its discreet location near one of the city's popular night spots belies the fact that trading arcades, like Kyte and proprietary shop Saxon Financial, are attracting intense attention from the financial services industry.

    These arcades, known as day-trading operations in the US, are racking up high volumes, spending big on new technology and drawing entrepreneurial IT programmers from software companies and banks' proprietary trading desks. They are notably employing algorithmic trading strategies in what is becoming known as "gray box" trading, which combines the work of humans and machines. Programs with pre-set parameters are executed with humans deciding the timing and direction of trades in response to market events.

    Though not a household name, when it comes to clearing futures and options in the UK, Kyte's transaction volumes are comparable to those of huge financial institutions in the US and Europe. The arcade trades more than 10 million contracts per month. Kyte has established itself as the largest clearing agent on London-based derivatives exchange Euronext.liffe and the third largest clearing agent for bund options on Frankfurt-based derivatives exchange Eurex.

    Investment in algorithmic trading has given arcades such as Kyte an increasingly large share of volumes on major exchanges, according to market participants. Trading volumes of Euronext.liffe's three-month Euribor futures contracts are now evenly split between banks and trading arcades. Euronext.liffe says algorithmic trading already accounts for about a fifth of overall volume on the Euribor contracts.

    "Our traders enter the market in the name of Kyte and we clear all their trades," says Peter Green, CEO at Kyte. "They are looking for a conduit to the exchanges and a safe place to leave their money. A daily statement provides an accurate representation of their open positions and the profit and loss they have generated in the last 24 hours."

    Set up as a stand-alone trading operation in 1985, Kyte has 400 traders who deal on their own account or as part of small trading teams. The trading arcade also offers lines of credit to traders to give them the capitalization they require to pursue certain trading strategies.

    Hyper-Fast Infrastructure

    Green and the group's founder, David Kyte, both dealt in the trading pits of the London International Financial Futures Exchange (Liffe) in the 1980s and 1990s. Yet unlike many of their contemporaries, who were unable to migrate from dealing floor to computer screen, Kyte and Green gauged the direction in which trading was heading.

    Kyte recently invested heavily in low-latency technology and algorithmic trading systems. Half of the contracts, which are traded at the arcade, are now generated by algorithms. "The technology is there to become faster and the style of trading among our client base is high-frequency where the speed of execution is absolutely crucial," says Green.

    Kyte has increased its IT spending from between 20 percent to 25 percent of the group's overall budget to about a third over the past three years. These resources have been used to develop the trading group's network infrastructure and triple the size of its datacenter, which houses blade server technology and CiscoKits, used to link internal and external servers to the trading system. "Kyte has created a stable and hyper-fast infrastructure so that when traders sitting at their desks send orders into the exchange, they pass through our network and on to the exchange in the fastest time possible," says Green. "One of our larger customers has a typical time frame of less than 60 seconds to be in and out of a trade."

    Kyte offers its clients proximity hosting at Eurex, whereby a server is installed next to the matching engine of the exchange to enhance trading speeds. This has reduced Kyte's roundtrip time for orders to 3 milliseconds. Given the sensitivity of algorithmic trading systems to market movements in price—which can leave behind those market-making decisions based upon obsolete data—access to such speed is crucial to Kyte's traders.

    "We know what the time savings are and they are not insignificant. A particular customer of ours is firing quoted prices into the exchange more times than every second. Sub-second prices are firing into the exchange on which other people can trade. The machine is over there in Frankfurt and the trading parameters are set over here [in London]," explains Green.

    The Arcade Advantage

    Christopher Morris, a director of London-based proprietary trading arcade Saxon Financial, argues that the trading venues are in a better position than the major investment banks to focus on the opportunities offered by algorithmic trading. "Smaller entrepreneurial organizations can be more successful than everyone else [in the current electronic trading environment]. You have to be able to shift gears quickly. Investment banks are held back by factors such as internal politics, signoff boundaries and the performance cycle," Morris says.

    Saxon, whose offices are located in a warren of medieval streets situated between the City and London's East End, draws traders with IT programming backgrounds on banks' proprietary trading desks. "We attract traders who want a fairer return for their intellectual property. Taking an entrepreneurial approach is critical to succeeding. Algorithmic trading is a moving target. A profitable opportunity one day may not exist the next. If you cannot change your focus, it is problematic," says Morris.

    Green says a handful of senior programmers from leading independent software vendors (ISVs) have joined Kyte over the past year to start new careers as traders. The programmers usually work alongside traders in small specialist teams of two to 12 people. Green says that the programmers are pioneering a "gray box" style of trading. "This style of trading has been widely used in the arcade sector over the past six months. It is something of a hybrid model between the traditional screen-based trader and a black box. Clever bits of programming are used but there is more human involvement than in a black box," says Green.

    However, Green says arcades aren't better poised than investment banks to exploit algorithmic trading opportunities. He points to the substantial investment that major firms have made in automated and algorithmic trading technology. Lehman Brothers trades $20 billion globally in equities each day, half of which is delivered via algorithmic trading platforms, according to Steve Vandermark, head of algorithmic trading and analytics at Lehman. The high volume of trading data that this activity generates makes the bank better placed to refine algorithms than those operating out of small-sized operations, he says.

    "If you look at Lehman in Europe, we execute through our algo platform about $5 billion a day in European equities and futures, and we leverage that data to back-test, refine and improve our algo implementations. Order-book dynamics, feedback effects and specific market structures are complicated and algo trading can only really be refined through extensive trading and extensive analysis of the data. The smaller firms are massively disadvantaged in that respect and this is a big reason why large banks should have superior performance versus the smaller organizations," says Vandermark.

    But Kyte's Green says the close-knit structure of trading arcades does allow the organizations to foster innovation in the algorithmic trading space. "Networks are designed only for traders and maybe we are more nimble [than investment banks]. If a new piece of technology is available on the market, we can get our hands on it quicker. We do not have to go through three committees to decide whether we are going to invest in this new product or not. There is live interaction, which results in a quicker time to market than in an investment bank where you may have a software development department that sits on the fifth floor while the traders are on the fourth."

    A trader on the proprietary desk of Société Générale says banks may be less willing to experiment with IT programmers because the algorithmic trading strategies they execute may take more time to become profitable. He says IT programmers recruited on the bank's prop desk from its IT department struggled to generate profitable returns quickly. "When you take into consideration the price of a trading seat—which consists of the fees the bank pays for trading and the overall running costs of operating the trading department—it is quite expensive to employ a programmer who may take a couple of years to be profitable. It is not only quantitative models that make money," he says.

    But Green says he is confident that Kyte's increasing focus on algorithmic trading is taking the arcade in the right direction. "Ten years ago, we had open-outcry traders who would trade by shouting and using a piece of card and a pencil. As the methods of trading have changed, we have moved with them. I do not see an over exposure to algorithmic trading as being detrimental to our business. We will just move with the times," he says.

    Joe Morgan
  2. snackly


    I am not so sure. When it comes to trading the banks are at a disadvantage and the reason is that hedge funds have multiple prime brokers and are able to discover their books and their flow through those relationships. The banks themselves do not share that data for obvious reasons. Small shops have a major advantage when trading, especially when it comes to currency.
  3. Haha is right - whatever happened to their Montreal office?
  4. I keep hearing about algos and algorithmic trading but have never seen an example of anything that is really sophisticated.

    Yes, brokers use algos to search dark pools and exchanges for liquidiity and place customer orders but I don't think this is something too sophisticated.

    I sense that people confuse order processing algos with trading system algos. This confusion is all over the place and it is starting becoming too irritating.

    Algos for order processing and execution will not make you money because they are not trading systems. That is the reason those morons have observed that IT professionals take 2 years to become profitable, at least some of them.

    It's because it takes them 2 years to understand the difference between information processing and trading system development.