Playing the Auto-Trading Game for the Mass Market
An up-and-coming vendor is hoping to tap into the burgeoning demand for automated trading.
Radial Systems, a two-man operation based in New York, is courting day traders, hedge funds, traditional asset managers and traditional broker dealers with a system that automates traders' strategies.
"We have developed a platform that enables us to take an algorithm or strategy and execute it automatically," explained Radial's Daniel Azoulay. "The system then manages the trade, manages the risk."
Auto-trading, also known as Black Box trading, is sweeping Wall Street at all levels. Day trading firms, bulge bracket prop desks, electronic desks, market makers, and electronic agency brokers are all turning to computers to trade their own books as well as client orders.
With the tremendous increase in the number of ticks and the flattening of spreads since the advent of decimalization, trading profitably has become much more difficult for the lone human. Auto-trading makes it possible for a single trader to analyze more data and handle more names. That allows his firm to keep costs down and/or make more money.
Auto-trading has traditionally been the milieu of sophisticated brokerages – big and small – trading for their own accounts and those of quantitatively-driven money managers. Bulge bracket firms employ Black Boxes on their prop desks. Proprietary day trading shops such as ATD and Tradebot/ Northtown auto-trade on ECNs. Highly sophisticated electronic agency brokerages such as Investment Technology Group push auto-trading out to their quant customers.
With the advent of decimalization, large market makers started deploying the technology in their Nasdaq dealing rooms to handle more symbols with fewer traders. Also, bulge bracket shops such as Morgan Stanley, CSFB, and Goldman Sachs set up electronic desks to facilitate their customers' simpler trades with automated algorithms.
In the past two years, auto-trading has spread to the broader market. In February 2002, market data filtering vendor Neovest added auto-trading to its line-up. "Due to the technology's ability to literally automate and execute tens of thousands of trades per hour," the company announced at the time, "the code was subjected to the most rigorous beta testing of any product feature in Neovest history."
Bruce Byers, president of Neovest, then went on to gush: "This technology will change the landscape of trading as we know it."
A year later, award-winning front-end developer and direct access brokerage TradeStation Securities released Version 7.0 of its TradeStation trading system, giving its day trader customers automated trading for the first time.
Mundane and Esoteric
Last summer, the proprietary day trading house Echo Trade, a unit of Chicago clearing firm Pax Clearing, introduced Predator Pro. The system purports to automate strategies both mundane and esoteric: opening orders, envelopes, baskets, gap finding, auto-gap entries, spreads, pairs and sector trading. Traders can use canned strategies or load their own. Once done, they can "sit back and let the computer trade for them," according to Echo.
The trend has even made its way into the halls of academia. Two years ago, the department of computer and information science at the University of Pennsylvania launched the Penn-Lehman Automated Trading Project. With financial and intellectual support from the quantitative trading group at Lehman Brothers, the project pits teams of traders against each other in a money-making competition. The project's goal is a "broad investigation of algorithms and strategies for automated trading," according to the university.
Despite the presence of in-house technology at Echo, Radial Systems has made inroads at the prop shop. A handful of Echo's 300 traders use Radial's system to automate their strategies.
Brian Ferdinand, in charge of Echo's New York office, explains that the assistance the computer gives traders is useful during quarterly earnings seasons. Traders can automate their traditional strategies while manually trading stocks subject to earnings announcements.
"This allows them to stay focused on earnings season," Ferdinand said, "while still implementing those other strategies."
Radial was founded by Azoulay and Amit Livnat in 2000 and is run out of a small room in Pax Clearing's New York offices. The vendor has made most of its initial sales to proprietary day trading brokerages such as Echo, Canada's Swift Trade and the now-defunct Worldco.
Proprietary day trading brokerages employ traders to trade the firm's capital. They are distinguished from "retail" day trading shops where traders use their own capital, but the firm's technology and offices. Prop traders split their profits with the house. Retail traders pay the house a commission.
Auto-trading, like all trading, can be boiled down to entry and exit strategies. That means when to put on or take off a position; how big of a position to trade; at what price to get in or out; and where to trade.
Once a position is put on, Radial's technology manages the trade in line with a trader's risk guidelines. That, in essence, means deciding how much money he can afford to lose in a day, either across his pad or on an individual stock basis. Loss parameters can be based on dollar amounts, say $5,000 per symbol, or the number of trades in that name. (e.g., "three loss-making trades and I'm out." )
Knowing when and how to enter and exit is dependent on a trader's assessment of market conditions. That assessment is boiled down to a number of if/then statements to form a strategy. For instance, if a stock has reached its high for the day, then maybe it's time to sell.
Trader decision-making is typically much more involved than that, of course, but "reading" the market is a key part of what Radial does. The technology sucks in every tick – every bid and offer and every sale – for every security being traded and then analyses the data and puts on the positions using the trader's if/then statements.
Testing, Azoulay says, is a crucial part of the process. Strategies are tested with paper money before, during and after the system goes live with traders' algorithms. There are four phases of testing.
Radial first tests the trader's algorithms when it is first presented with the idea. Then, once they know the strategy is working, the partners perform tests with different configurations over several days. The strategy is "optimized" to ensure the best performance.
Then, once the trader has put the system to work with real money, Radial tests alongside the trader, but with paper money. That's to see whether or not the strategy is working as intended. It may not be.
The one bugaboo with all automated trading systems, Livnat explains, is "slippage." Because it takes time to hit a bid or offer or place a limit order, the execution may not occur as expected. Market impact also plays a role.
"Most of the Black Boxes on the market are failing with this issue," Livnat said of slippage. "They see good results from the paper trading, then start to use real money and lose." Radial has developed a technology that allows it to estimate slippage, according to Livnat. "By testing in parallel to real trading, we are able to constantly verify that our technology is matching the live trading," the developer says.
Finally, Radial will tweak its programs by running simultaneous tests after the market closes. Because it has access to Echo Trade's machines, Radial can test on several computers at once, allowing it to run a strategy with a myriad of different parameters. Using multiple PCs saves time.
Variables tested could be the amount of a stop loss order; the length of time a position is held; or whether to stop trading at 3:40 p.m. or 3:50 p.m., for instance. "There are very few players out there testing hundreds of different configurations to optimize their models," Azoulay said. "We do. Not only are we going to program it for you, we will help you make it better."
Sophisticated Traders
Auto-trading is not for everyone. Generally, only the most successful and sophisticated traders use it. Within the ranks of day traders, however, it is the successful ones who were left standing following the market rout and the move to decimalization four years ago. These were the traders, says Azoulay, who use a consistent methodology to trade.
"A trader with a successful track record is usually doing the same thing or something similar every day," the exec said. "No one is going to trade differently every day and make money consistently. They have a way that they trade. They look for signals."
Those signals may work for hundreds or thousands of symbols, but the lone trader can only deal with a handful at a time. "Traders know they miss hundreds of opportunities because they can't handle it," Azoulay said. "With our system, instead of doing it for five symbols at a time, you can do it for 5,000."
