Printer Friendly
When markets are volatile, using algorithms -- especially those that are based on historical data -- brings increased risk. Who wants to throw a large order into an algorithm only to watch the trade get away from you because the market is reacting differently than the algorithm could predict based on the market's past performance? As the markets become more and more unpredictable, buy-side institutions increasingly are handing over large or difficult trades to a person who has the time and expertise to break up the order and work it without moving the market while adjusting the trade as the market changes. Research and strategic advisory firm TABB Group says that the use of the sales trader has increased from 37 percent in 2007 to 44 percent of volume in 2008.
Nonetheless, in a discussion back in mid-December, Robert Kissell, head of quantitative trading strategies with J.P. Morgan, told me that even during these volatile times, he was experiencing a high demand for algorithms. He noted, however, that clients are changing the "settings" they use with algos to trade much more aggressively. Most algorithms, Kissell explained, have three settings: a low level of urgency, a mid level and a high level. Most traders, he said, are shifting their strategies from low urgency to high urgency in order to get trades done in a matter of hours rather than risking a major market swing while trying to trade over the span of a day.
Still, J.P. Morgan, like other firms, is taking the opportunity to rewrite or upgrade algorithms to reflect the changes in the market, according to Kissell. "We want an adaptation strategy to take advantage of market conditions when they arise in our favor and when the market is working against us. We need to be able to incorporate this information to better manage our trading risk," he explained. "We have rewritten a lot of our programs to look at market prices and where the market has moved since the algorithm started trading."
Kissell added that many firms are trying to address volatility by creating volatility estimates and incorporating them into their algos. But, he acknowledged, this isn't always easy, especially with some options and small-to-mid-cap stocks.
This all is particularly important when it comes to using algos to execute block trades. As a result, in many cases the human has come back into the equation. (See "Volatility Shakes Up Block Trading".) Despite this, Kissell said, he believes that the use of algos will remain very strong, mostly because they are such an efficient way to trade.
Emphasizing the importance of building algos that adapt to this crazy market, Kissell added, "The only thing I am certain of is that the market conditions are going to be different than what I planned for in my initial optimization."
