Have you read I ROBOT?
Algorithm Developers Go Back To Basics, Enter Options Mkt
.
By Mohammed Hadi
Of DOW JONES NEWSWIRES
NEW YORK (Dow Jones)--The developers of electronic trading software are going back to basics to give their customers options, stock options that is.
Trading algorithms are software programs which aim to mimic human traders, ideally with more efficiency and at a lower cost. These programs can be used directly by hedge-fund and mutual fund managers to lower trading costs and handle mundane jobs so that human traders can focus on more complicated orders.
Now, as algorithm developers seek new venues for their products, these programs are being rebuilt to operate in the options market. At the moment, the developers are still working with basic building blocks of what will eventually become sophisticated algorithms, called smart order routers.
"Smart order routing is just taking hold as the new secret weapon in options trading," said Rob Flatley managing director of electronic trading services at Banc of America.
Because fund managers are already familiar with the technology, software developers think the growth of the new options algorithms is likely to be much faster than that of equities.
"This is history repeating itself," Flatley said, "the difference is that the velocity of change is much greater."
That's simply because fund managers who already use algorithms for their equity trades don't have to be "converted" to the technology. "We didn't have to do a lot of missionary work on selling them the benefits of electronic trading," Flatley said.
Flatley and others say development of options-related algorithms has only come about now because of the continuous shift towards electronic trading, which began when the International Securities Exchange, or ISE, launched in May 2000.
To remain competitive with the ISE, which quickly gobbled up market share, the other options exchanges have also taken steps to boost electronic options trading.
At the same time, regulatory changes and the removal of certain hurdles by the exchanges have made them a more welcoming place for institutional investors and, therefore, trading algorithms.
"The options markets are now tight, efficient, competitive and highly liquid," said Kim Bang, president and chief executive of Bloomberg Tradebook which launched an electronic options trading platform last month.
Plus, brokerages which have already poured millions into research and development of stock-trading algorithms are no longer operating uncharted waters, at least not where the basics are concerned.
Goldman Sachs, for example, now offers a version of its smart order routing software, called Sigma, for options trading. "The program breaks large orders up into smaller pieces and then sends those pieces simultaneously to the different options exchanges where smaller matching orders sit," said Greg Tusar, Head of U.S. Electronic Trading.
Goldman's clients can access the platform from the firm's direct market access platform, called REDIPlus, and daily volume of options traded on that has almost doubled since last year.
But in the last six months Goldman has taken things one step further with a new development called Prowler, which is a type of order that can be placed with the Sigma router. Prowler also breaks up large orders, but instead of filling them wherever possible, it holds back and "only sends the piece that can be executed electronically based on the display size," Tusar said.
"It's the first real algorithm that we've developed to be more like what we do in equities," he said.
Indeed, one of the benefits touted by algorithm developers is that these programs can minimize information leakage - that is keep the market from knowing that someone has a large order to fill - and therefore minimize the impact that a large order might have on a stock -- or in this case an options -- price. Orders filled by Sigma or Prowler currently account for nearly a quarter of the 600,000-or-so options trades that take place REDIPlus each day.
Still, even in the world of equity trading, where they are fairly well established, algorithms only account for a small portion of total trading. Flatley estimates that by 2007 they will account for 20% of all equity trading, up from about 5% today, and many fund managers keep their use of this software limited to only the most easy to trade stocks, saying that their value diminishes in difficult situations.
Things are only more complicated with options trading.
For starters, the most advanced stock algorithms consider volumes of data - both historical and real time - when determining the likely outcome of a trading decision. That data comes not just from the stock market, but can also come from other markets that might influence how a particular stock trades. When it comes to options trading, data can be more of a problem.
Because there are multiple options listed on every stock, the data that options algorithms have to process is several times greater than that of stocks, and according to the number of messages being transmitted has more than doubled in the past year, to as much as 80,000 messages per minute.
Options brokerages and data providers do complain that keeping up is a costly effort that requires the constant upgrade of technology.
And the algorithm developers whose businesses are already involved in options trading also say that this actually gives them an advantage over competitors who have flooded the market for equity algorithms.
For example, Interactive Brokers, which offers smart order routers for options trades is part of the same company that owns Timber Hill, an options market making firm. As a result, the firm already bears the cost of keeping up with growing quote traffic, said Steve Sanders, managing director for business development at Interactive Brokers.
"Its an expense necessary for us or we'll be out of business," he said.
Dealing with quote traffic, Flatley says is a real focus of research at Banc of America. Even if rivals can pay for the technology used to handle the traffic, "the real black magic in this is how you compress that data enough to make it move bullet fast across that network," he said.
Still, there's more to it than technology constraints. Unlike stocks, which are issued by companies, options don't exist in nature. They are contracts that have to be created by an agreement between a buyer and a seller, and more importantly, they expire. That means that liquidity in the options market is far less predictable than it is in stocks.
It also means that historical trading patterns are difficult to establish, and, for stocks that rise or fall quickly, old contracts can lose their relevance.
Algorithm developers acknowledge that this is a limit, saying that for now they know electronic trading will remain focused on the most predictably liquid issues, like options on indexes.
"Option orders are increasingly executed electronically but you wouldn't leave the entire order to be executed by a machine yet," Tusar said.
Plus, that makes things more complicated than just copying existing stock based algorithms. "Its a little bit different in terms of being opportunist and accelerating or decelerating the model based on available liquidity," Flatley said. |
