Systematic trading is the ticket.

When Futures asked me to discuss the evolution of my trading strategies over the years to what they are today, it gave me the excuse to reminisce about years of personal development. With numerous acquaintances in the CTA community, I've been able to observe the development of many other traders,

both discretionary and systematic, in our industry. It doesn't make a difference whether trend followers are "in" or "out" with the asset allocators in a particular year. I admit to being a systematic long-term trend follower, and I remain a strong supporter of systematic trend-following trading techniques.

Most of the good returns for the industry come from those times when there are large volatility and trends. Usually, during these highly profitable periods, you'll find all traders, discretionary or systematic, having their best months. For anyone to make money in a market, there has to be a trend to exploit.

Trading is managing information flow

When you break down everything to its smallest function, every trader in the futures industry spends most of his/her time managing information flow. It matters not whether the trader is a fundamentalist or a technician; the trader could believe in Elliott wave, Gann or see some other form of order in the universe. The trader could rely on government reports, analyst's opinions or the advice of a friend. Whatever the approach, information processing is at the heart of it.

So what's the most efficient way to manage information flow? The answer is to computerize it, of course. That now puts you squarely into the realm of the systematic trader and is the major reason why Trendstat is systematic.

Before you computerize your approach...

The first issue you have to deal with before automating your trading strategy is why you are in the business of trading. Are you simply trying to manage client assets efficiently? Are you trading for the thrill of trading? Are you trading for the money you hope to make from successfully trading, or do you love the challenge of solving the riddle of trading well?

If you are seeking any thrill or excitement from trading, you'll have a hard time becoming a systematic trader. You may automate pieces of trading, then overlay some discretion on the system's decision. You may invest with more leverage if you feel good about a signal or stay a little longer in a position that is not hurting you badly. In my own development, when I concluded that I didn't need or want the excitement of trading, I had arrived at a good mental state to automate my trading style.

Our strategic definition of Trendstat, which should work for most traders is: "We are an information processing firm that manages investments for clients in the liquid markets." Nowhere in that definition is "money manager" or "CTA." We are information processing specialists that happen to focus our information processing technology toward the management of investments. What a difference this focus gives you. You find yourself asking "What information is needed to make this type of decision and how can we obtain it?" "How do we get it into the computer and have it do something with the information?" You do not sit there all day glancing at a quote screen, worrying whether or not your stops will get hit. There are days that I don't even look at individual positions because my computers are watching the markets for me and I trust them to do that better than I could myself.

Incorporating your strategy

There is little difference between good discretionary trading and good systems trading. The good discretionary trader has given some thought to a general strategy that will be used and executes it every moment with a degree of discipline. The good systems trader uses discretion to map out a strategy to trade and uses a computer to execute the system each day. Both use some long-term discretion and some short-term discipline of some sort. To me, good trading is good trading.

What makes the markets work?

My personality is long-term and patient. I was a chemical engineer by original degree, so I have a good working knowledge of computers and math. So when it came time to create our trading systems, I first developed a philosophy of how the markets work. I like simple. In my simple view of the markets, buy orders create upside forces or demand. Sell orders create supply or tend to push prices lower. The battle of the supply and demand of orders goes on all day, with the winner ultimately setting higher highs or lower lows.

I started out in the 1970s with sheets of graph paper and was constantly changing everything I did. It took me about four to five years of trading to evolve to strategies and discipline that allowed me to make a small profit. As more disciplined trading became part of my day, I realized I was spending a lot of time performing what I considered repetitive, mundane tasks. These included, but were not limited to graphing data, generating buy and sell levels, deciding on how many contracts to buy or sell and calling in the trades. I decided, for quality of life reasons and efficiency, to give the computer the job of doing my menial work each day. I've found that having done that, I now have more time to do things that human beings are uniquely qualified to do, such as creating new strategies, analyzing old strategies, helping clients and writing magazine articles.

Develop a trading philosophy

The development of our trading systems flows as shown in "Levels of system development." With a good foundation of how I believe the markets move up and down shown at level one, I then develop a philosophy of how I'd like to trade the markets. Right now, Trendstat has two trading strategies and a third is under development. Each one has different missions. Each has a different type of market it is trying to exploit and each has its own risk profile.

Let's develop a simple trading system as an example. At level two on the flow chart, we need to develop a trading strategy philosophy. Let's say that we want to let our profits run and cut our losses short. We'd like to pick up every major move of the markets and not trade every other day. This should keep our trading costs down and minimize slippage costs. We are willing to take a few losses during sideways markets but are not willing to take these losses every day. We would always like to have a stop to protect us against adverse market moves. These are all the types of attributes I'd develop at level two.

Define the indicator(s) to fit the philosophy

The levels keep building on each other. At level three, we need to develop a specific strategy that fits the profile we've described in level two that fits our view of what makes markets go up and down. A moving average wouldn't work, because it might whipsaw in sideways markets every other day, and we didn't want to do that much trading. Overbought/oversold indicators like RSI and oscillators would not be able to pick up every major move. Fundamental indicators get too far away from that which is going to make or lose money: the prices of the market we are going to trade.

One style of indicator that would fit our philosophical profile would be a breakout indicator. We would get a signal for every major move. Sideways markets would undoubtedly whipsaw us but not every other day. Profits could run, and losses would be defined. Therefore, at level three, we set to work developing the math to describe a range or channel breakout indicator.

Without giving away the store, our indicators are a bit difficult to describe in a short article, so let me create a simple model that will illustrate the point. Let's look at using a simple 30-day breakout indicator. The math for a signal would be:

Buy if price is greater than the maximum high of the previous 30 days.

Sell if price is lower than the minimum low of the previous 30 days.

"30-day breakout" (next page) shows a simple graph of where the buy and sell signals would be with this simple strategy.

At this point in this simplified process, we've completed level three and have the foundation to go to level four. We would spend some time programming our computers to take the price, look at the last 30 days and tell us where our stop orders to buy or sell should be. We can program it to look at as many markets as we care to trade. Trendstat looks at about 20 liquid markets. It would be just as easy to look at 60, although some of these markets might be giving you more of the same thing or be somewhat illiquid.

Historical simulations

The final step is to analyze the model's results. Here is where I would differ with many systems designers. I don't attempt to go back and get years and years of data believing that if I can get enough good data, I'll be able to rigorously test the model and give it a passing or failing grade. Many system designers do just that, and I believe they fall to realize that they are not preparing themselves for the eventual event that falls outside the data they used in the simulation. They would say if they got enough data, they should be okay. I would say that 100 years of stock data would not have included a Crash of '87.

So how do you properly review historical simulations? I approach it from the levels in "Levels of system development." I know the philosophy of the trading strategy was to make larger profits in major trends and probably have difficulty in sideways markets. I observe the details on the simulated track record and see if the results for major trending periods show good profits. Does the track record show losses during sideways periods? What periods were the most difficult for the model and what was happening in the markets during that period? Where was the best period and why did the model react so well during that period? How much of the profits are from a few good trades? What is the sample size of the trades and what reliability did it show over the period?

This way of reviewing historical simulations is extraordinarily time-consuming. I've been known to sit with a computer printout and a ruler for two or three hours observing each buy and sell signal, why it occurred and how much risk did the simulation expose the portfolio to each day of the period tested. Looking at some summary that comes off the end of many of the popular software development packages doesn't do it for me. I need to see the detail to know that the system is acting like I expected it to.

Many system designers make the mistake of jumping up to level four right away, creating or using a software package to simulate some trading, then looking at the results as good or bad. If the results look okay to trade, off they go to the markets to trade their new system. Inevitably, the system runs into a rough period, they get nervous. decide the system isn't working anymore and abandon it, only to turn around and repeat the same ridiculous process of systems development again.

Trading in the future

My own obviously biased view of where we are going is toward more automation. We shouldn't fight it. We should welcome it as liberating us from the mundane, information-processing tasks associated with our lives. This allows us the time to create, think. relate and help. Computers will definitely be more a part of your life in the future than they are today. Systematic traders will be more able to automate their process than other trading styles. I look forward to tomorrow's more automated world.

RELATED ARTICLE: Levels of system development

5 Evaluate the program on how well it matches your philosophy. 4 Build a computer program that is based on logic below. 3 Select a general type of math or logic that matches philosophy. 2 Develop a philosophy toward exploiting the markets. 1 Supply and demand moves prices up and down.

Tom Basso is president of Trendstat Capital Management of Scottsdale, Ariz. Trendstat is a registered commodity trading advisor and investment advisor. It is currently managing approximately $115 million.