Trading systems once were reserved for institutional traders with access to powerful mainframe computers and Fortran programmers. Things weren’t all rosy even for these well-heeled institutions, however. Market data was not disseminated as neatly as it is today. Usually, it was necessary for a paid assistant to spend hours entering data manually from newspaper quote pages into a database.
When personal computers were developed in the early 1980s, the seeds of system trading as we now know it were planted. Things really took off in 1989 with the birth of SystemWriter from Omega Research. Well-known traders like Larry Williams made names for themselves using this product to research and develop trading patterns and systems. The movement really took off in the 1990s when SystemWriter evolved into the Windows-based TradeStation, which supported systems-based day-trading.
Simple systems such as opening-range breakouts and channel breakouts were money machines in those heydays. Today, however, much of the luster has worn off these techniques, leaving many traders to wonder whether the earlier success was simply a product of chance or whether the spread of these computerized systems somehow dulled their effectiveness.
For traders today, this is much more than an exercise in historical accuracy. It can help us understand how current changes in the markets might affect the systems you are using to trade today.
TRADING SYSTEM TECH
One description of markets is that they are mechanisms designed to take money from the many and redistribute it to the few. Cynical, yes, but it’s also a description that bears out in empirical evidence, and when “everyone” started using systems from the same family, “everyone” couldn’t win. As a result, those systems became less effective.
Procedural changes in the markets themselves also can affect the performance of trading systems. One memorable example is in the Treasury bond market. Before 1988, T-bonds opened at 8 a.m. (central) and financial reports were released at 7:30 a.m. This allowed the report information to be processed and the market to move in the expected direction off of the open. As a result, the opening-range breakout system was a virtual money machine in T-bonds.
Before this change, Williams and Sheldon Knight, another well-known trader, were able to turn relatively small accounts into more than $1 million in the mid-1980s. In 1988, though, the fun was over. The open time was changed to 7:20 a.m. and the period immediately following the open was dominated by noise while the report was digested.
Beyond client-designed trading systems, SystemWriter and TradeStation were responsible for another boom in trading systems: They allowed system developers to build and sell indicators and systems to other traders easily. The birth of third party add-on programs further fed the system trading revolution.
This, in turn, spawned other types of businesses. Some tracked the commercial trading systems. One of these, if not the first, was Futures Truth, which was founded in 1985. In 1987, the company started publishing a report that tracked commercial trading systems and even developed a software package, called Excalibur, to program and backtest trading systems. Another related service was The Club 3000 Newsletter, published by Bo Thunman. Effectively a printed blog, the newsletter shared the experiences of end users with the various trading systems they had purchased and traded.
As expected, the majority of commercially available trading systems did not work well. However, some did persevere, spending many years at the top of Futures Truth’s independent rankings. Some of these, such as Rick Saidenberg’s R-Breaker and Keith Fitschen’s Aberration, sold many copies.
As traders discovered what systems worked and what didn’t, they started to figure out another hurdle to trading success: the difficulties of managing a trading program while working a day job. This time, the market evolved to provide traders automatic execution and management of their systems. One of the pioneering companies here was Robbins Trading Co., where a broker named Joe Krutsinger played a key role in developing a service known as System Assist.
Other brokers followed suit, but it wasn’t until the advent of electronic trading when this business, and system trading in general, really exploded.
It was slow going in the early days of electronic trading. As recently as 1997 and 1998, volume on Globex in the S&P 500 and on Project A in Treasury futures was relatively low, and traders had to worry about being stopped out during the night by an outlier trade.
These early concerns, and others, inspired studies of the unique dynamics of electronic trading. One of the most important was written by Satu S. Parikh of the Wharton School of Business who looked at electronic futures markets vs. floor trading.
This paper discussed how pit-based, open-outcry traders could share information above and beyond the process of simply placing orders, which helped to create efficient markets. Indeed, from body language to institutional knowledge of who filled orders for whom, and how those orders might roll out into the market, seemingly ancillary information often conveyed more data than orders themselves. Parikh’s concern was how all this would be lost with a move to electronic markets. Parikh’s findings had major implications for interface design, which quickly gravitated toward a market-depth display that not only showed the current spread, but orders “in the book” waiting to be executed. The exchanges’ decision to offer market depth advanced electronic trading by providing something floor traders didn’t have, at least not definitively, and spurred innovation among front-end providers.
This design was important because the bread and butter of floor traders in the futures markets is trading the bid/ask spread. This adds necessary liquidity to the markets. Without these floor traders, slippage could become so great that computerized trading systems using stop orders would become untradable.
These were major concerns in 1995 and are even more important today with most markets now traded electronically. Also thanks to the preponderance of low-cost online brokers, off-the-floor day-traders are no longer an exception. They are the rule.
TRADING SYSTEM FALLOUT
If a minor shift in the opening of T-bonds can have a major effect on one of the more popular means of trading that market, what will happen when we get rid of the concept of an open altogether? For that matter, what, exactly, will the open be? While it’s not perfect for futures traders, in two markets, bonds and stock indexes, extremely liquid cash markets help sort out the existence of a real opening price. In other markets, such as grains and energies, there is no structured and active physical cash market to provide this proxy.
Another consideration is the effect that 24-hour electronic trading has on stop orders. These orders are used to enter and exit positions and are often hit during off-peak hours. While this is allayed somewhat by the influx of orders from overseas traders, the overnight volume is nowhere near the numbers generated during the day.
We are truly in an era of the unknown. Once 24-hour electronic markets have replaced pit-trading sessions completely, many systems will change. The effects will be greater on shorter-term swing systems than on longer-term trend following systems. The good news is that all of this change won’t necessarily be negative.
One effect is that even with stop orders, many markets during the day session hours have slippage of no more than one or two ticks. At one time during 2001 to 2003 in the stock index futures, realistic slippage was from $300 to $500 on a big S&P contract. Because electronic trading offers low commissions and reduced slippage, the realistic cost of slippage and commission can be as low as $30 to $40 on an E-mini S&P 500 contract, even in volatile market conditions.
This is also true of other markets, such as energies and grains. Systems that make as little as $100 per trade before slippage and commissions that were useless a decade ago can now be re-evaluated, thanks to electronic trading.
HOW TO ADAPT
Clearly, the biggest change will be the effective loss of the day session. Traders and trading systems are well-tuned to the concept of day and overnight sessions. We’ll all have to change when 24-hour trading becomes a reality.
One of the biggest effects will be on trading systems developed on end-of-day data. Traders will have to face tough solutions, such as whether to re-build those systems, to trade them on 24-hour bars or to create artificial day sessions. We could, for example, build a price bar using futures data from 9:30 a.m. to 4:15 p.m. for the E-mini S&P 500 to mimic traditional day-session data. However, this would only be remotely feasible for futures markets with large, liquid cash markets, such as stock indexes and bonds, to back them up. Even then, some might argue this exercise would be akin to using a buggy whip to encourage a Model T down the road.
Some forex traders believe that they have found a solution. Twenty-four hour trading in the cash forex market has been a reality for several years. Forex educator and commodity trading advisor (CTA) Osman Ghandour trades a program using the old 4 p.m. (Eastern) close for foreign exchange. While arguably arbitrary, the time represents the end of peak liquidity and, for forex, is preferable to relying on the day trading hours for currency futures.
Another possible solution would be to set times based on volume. A “day session” could become an “active session.” For this to work, some consistency of volume patterns would have to be present, and that might not be the case for markets impacted by business cycles or agricultural seasonalities.
To better understand the importance of these considerations, look at the performance of a basic 20-bar channel breakout system. We’ll look at the performance of both the day and the combined sessions on 30-year T-bonds and start the test on Jan. 4, 1998 because that’s when the night session in bonds became active. This test ends on Feb. 11, 2008. There’s no deduction for slippage and commissions because we’re only examining differences here, and we’re not necessarily interested in the absolute performance of either system (see “Overnight blues”).
Unfortunately, there are no easy answers on how to deal with the changes in the trading landscape, and as with all things in trading we probably won’t have that insight until we’re blessed with the benefit of hindsight. One thing is safe to say, however, and that’s that all trading systems will need to become more adaptive to be effective going forward.
Second, as a matter of course, once the markets are changed to electronic and trading histories have been combined, responsible traders should re-backtest their systems, assuming there is enough data to obtain statistically valid samples. Robust testing methods such as walk-forward analysis will become mandatory in the development of reliable systems. New areas of research will likely involve evolving parameters over time or even automatic parameter selection.
Even after we’re operating in a brave, new, fully electronic 24-hour world, tomorrow’s systems will need to adapt as those markets continue to mature and globalize. And expect some potholes along the way. The markets didn’t bring us to this new landscape overnight, and we’re not going to learn how to trade here that quickly either.
Murray A. Ruggiero Jr. is a consultant in East Haven, Conn. His firm, Ruggiero Associates, develops market-timing systems. He is vice president of research and development for TradersStudio and author of Cybernetic Trading Strategies (John Wiley & Sons). E-mail him at firstname.lastname@example.org