London black-box trading?

Discussion in 'Trading' started by walter4, Apr 22, 2007.

  1. Article on London black-box trading and more:

    "While market gossip may be behind much of today's movements, the effects of computerised trading may be helping rumours stay alive, with tittle tattle spreading through streams of bits and bytes.

    "We've not quite got to the stage where the computer programs are starting rumours," says one trader. "But some of the systems out there will pick up on any increase in trading volumes that come about as a result of a half-baked rumour.

    If those programs start buying, then other programs that follow prices might start to flag it up. If you're not careful, you can find that the computer programs have been trading against each other."

    There are a number of different computerised trading systems. The original systems, which first came into use in the mid to late 1990s, were simply ways of breaking up big trades.

    If someone wanted to sell 50,000 shares in BP, the software would break it up into 10 parcels of 5,000 shares and release them into the market during the day. By doing this, the trader avoided moving the share price too much.

    As new technology has increased trading speeds and the costs of dealing have been reduced, computer systems have been able to break trades into even smaller parcels. Different ways of disguising big trades have developed, using ever more complex systems.

    Anyone who wants to deal in big quantities of shares in a company like Vodafone or GlaxoSmithkline needs to use these systems now if they want to get the real market price.

    Dresdner Kleinwort is trialling a system called Stealth, which it believes can buy or sell shares in small, illiquid companies without anyone noticing. Credit Suisse has a similar system called Guerilla. Goldman Sachs has a system called Sonar, which finds spikes in trading volumes to smuggle through trades.

    "Sniffer" software is being used to find other software at work in the market in the hope that it will pick up trading opportunities. Getting the best price in the market has become a technology war.

    "Algorithms are certainly helping to increase trading volumes overall," says Eli Lederman, the head of electronic trading at Morgan Stanley. " Algorithms have changed the market structure - now you need to use algorithms to negotiate the new marketplace.

    Spreads have narrowed, commissions have come down and high frequency market dynamics have taken over. Trading desks would have to be staffed at outrageous levels to deal as much and as quickly as one has to now."

    The volume of trading by black-box systems in the UK market has tripled since the middle of 2005, according to Goldman Sachs. The effect has been remarkable. With the computer systems disguising all the big trades in all the big stocks, the volatility of shares in the FTSE has tumbled to historic lows. So how can the black boxes be to blame for spiking share prices?

    For one thing, the increased volumes being generated by computerised trades allows human traders to behave more speculatively. A trader can take up a speculative position for a few minutes, and know that he will be able to get out again. But it does not end there.

    "There is so little volatility in the market that there are very few trading opportunities," says one specialist in computerised trading at a major investment bank.

    "That means it is becoming ever more difficult to make money. So as soon as something starts to move, everyone piles in. Machines can overreact in that situation - the bad programs do, anyway. In some circumstances people then end up overreacting to the machines, thinking that there is something going on. "

    Some black-box systems are better than others. The simplest cannot be stopped from buying shares in a company once an order has been placed. Other systems are being used incorrectly, which is potentially leading to inaccurate pricing in the market.

    Says Peter Sheridan, head of European algorithmic trading at Goldman Sachs: "The growth of demand for algorithms has been so fast that sometimes people are not using them in the best way.

    There are some securities that are just not suitable for certain types of algorithm. There are also some participants in the market using less sophisticated, commoditised systems which do not function in line with the subtleties of the UK market.

    "For example, many algorithms being used in the UK market are programmed to trade higher volumes to coincide with the opening of Wall Street. Yet many UK stocks have a low correlation with the US market, and therefore will not necessarily exhibit a predictable volume profile. That's particularly true for mid-cap securities."

    When trading volumes are low across the market the capacity for the trading programs to mis-read the market is exaggerated. Many of the systems at use are not programmed to take account of the lower trading volumes around holiday weekends, for example.

    Experts are expecting the upcoming May bank holiday to show further evidence of the poorer systems misreading the markets. If the algorithms being used to place big chunks of shares begin to get things wrong, that in turn causes other electronic funds to make mistakes.

    A considerable number of hedge funds make their money by using computer programs to throw up trading ideas. There are two common strategies.

    The first is "mean reversion" - the theory is that if a stock rises or falls by a large percentage in a short space of time, it is likely to begin to head in the opposite direction. The second is "momentum trading" - the idea here is to start buying any share within milliseconds of it starting to rise.

    Has momentum trading been fuelling bid speculation surrounding companies like Prudential by buying at the first whiff of a price rise?

    If big chunks of money start to follow an idea, much of the market has no option but to follow. Roughly 30 per cent of the money invested in the FTSE100 is held through tracker funds - investment funds which are judged on how closely they follow the index.

    They use software to get as close as possible to every share's closing price at the end of the trading session - which often causes spikes in trading volumes late on the day.

    If a tracker fund is attempting to invest new funds received from clients on the same day as nonsensical takeover speculation is driving the FTSE, it has no option but to join in.

    While computerised trading cannot be blamed for all the froth in the markets at the moment, it is certainly a factor."