Heuristic-Based Trading
Although I am engaged in developing two algorithmic trading systems, the Trend Machine (a cellular automata?based system) and JOSTAN FX (an artificial intelligence expert system), I don?t use automated algorithmic systems to trade currently.
I remain an old-fashioned discretionary, seat-of-my-pants chart trader.
I use nothing but charts to trade, no indicators. I do troll the Global-View forum two or three times a week to get a sense of what other traders are thinking on specific markets, to see if I am missing something big, to learn about the fundamental underpinnings, and to scout for the possible contrary opinion opportunity. My trading is not mechanical. I have traded for 35 years and studied tens of thousands of charts. I firmly believe in keeping
it simple (KIS). While I seldom go against my trade plan, I am not hesitant to trust my instincts from time to time. Nonmechanical methods are less prone to the occasional price bluff than are algorithmic systems. The idea of 25-year-old quants with functionally no market experience running tens of billions of dollars in FOREX quite simply makes me shudder. A debacle is coming?when, I do not know.
I don?t trade the news, but note all announcements pending for the currencies I trade. I watch to see how the market reacts to these as an indication of the major trend. For an excellent analysis of the quantitative aspects of news announcements, see Forex Shockwave Analysis, by James L. Bickford (McGraw-Hill, 2007).
THE SNOWFLAKE TRADING HEURISTIC
Critical to my trading is the heuristic I have developed over the years. I call it the Snowflake heuristic and consider any trading heuristic that is process based in this manner to be a Snowflake approach. The idea of a Snowflake heuristic is for the trader to gradually zero in on a trade, working from the most general considerations to the most specific. For more on my thinking in this regard, see Getting Started in Forex Trading Strategies (John Wiley & Sons, 2007), where much of this methodology is described as the Codex process.
?Boxing a trade? was Mr. Goodman?s phrase for the process of finding a trade?going from a general observation of a market to a more specific analysis and finally to actually entering an order. I think ?Snowflake? describes it and my additions to it more accurately.
1. Analyze the market.
2. Identify candidate trades.
3. Determine money management parameters.
4. Enter the trade.
5. Monitor the trade.
6. Exit the trade.
7. Postmortem the trade.
8. Analyze the market.
The postmortem is an important step. I have found that small, incremental changes to my trade process can mean a large difference on the bottom line. But once my postmortem is finished, I do move on and do not think about the trade itself again.
Much of my trading method is based on two methods I learned from my mentor, Charles B. Goodman, and further developed on my own over the years since his passing in 1984. (The first two chapters in Part Three??Currency Futures Trading Basics? and ?FOREX Lessons from Shanghai BC??explain this in greater detail.) I have refined Charlie?s approach and customized it to my own trading style and personality.
The Snowflake trading heuristic is quite simple. As an eclectic, I am not a slave to any methods or ideas. I want to spend the majority of my time watching, questioning, and analyzing markets. I do try to follow the Snowflake heuristic to some degree for every trade. I am not by nature an organized person, so it requires real effort for me. You begin
with a general shape for a trade and continuously overlay it with heuristic steps until the shape is clearly defined and you have a trade. This idea was first introduced to me by Randy Ingermanson as an approach to writing. His Snowflake Principle obviously has applications to many domains. Play this short Java applet to get a visual sense of Snowflake:
http://math.rice.edu/?lanius/frac/koch/koch.html.
I am generally a day trader, although I have been known to be a scalper and to use scalping as a method of testing the waters and building a position in a pair. I monitor 10 to 15 markets at any given time. They are said to be ?in the hopper.? The pairs and crosses in the hopper are not fixed; I sometimes rotate currencies in or out as I see things on charts or as the spirit moves me. I review charts of 25 markets?5-minute, 15-minute, 30-minute, 1-hour, 3-hour, 1-day, 1-week?on a weekly basis. If a chart catches my eye or I read something interesting in the Global-View forum, I might make hopper changes.
Of those 10 to 15 markets, typically seven or eight will be on a watch list for which I keep basic notes. Three or four will be candidates that I observe and analyze more closely for Goodman Swing Count System (GSCS) trading formations, and one or two will actually be trades in progress. I rarely have more than three open positions at one time. Even with a relatively simple heuristic and trading method I have used for decades, it is still a lot to keep track of for me. Certainly one advantage of the algorithmic trading system method is that a computer can follow more markets than any one person can follow.
There are four steps to the primary trade (Snowflake) heuristic: Watch, Hopper, Candidate, and Trade. Prospective trades move from the former to the latter. Of course, many potential trades are eliminated during the process. The heuristic approach can also be useful for money management and attitude, although the trade heuristic is most critical. All of these are covered in the sections that follow.
Before examining the heuristic in more detail, here are a few tips to consider along the way.
? Unless you use an automated trading system, you are constantly juggling and judging a variety of inputs from the market. Being on for trading is essentially having all these elements working together smoothly. I take 10 minutes before each session to try to get those elements working smoothly in my mind and catch up on the market action that I have missed. I take 10 minutes after each session to take a mental snapshot of at least my current candidate trades and open trades. I never leave a session without stop-loss and a take-profit prices sitting in the market.
? Give a trade time to work. Most new traders are in a rush and spend too much of their time watching the shorter time frame charts. Remember, even in such active markets as FOREX it takes time for a trend to develop. If you don?t believe me, look at a few daily and weekly charts. You?ll see some enormous trends?but they took days, weeks, and months to develop.
? Do not let a profitable trade turn into a loser, ever! You can always reenter a market. ?If in doubt, stay out!? Assuming you are a day trader, once a trade has gone 20 to 25 pips in your favor, consider taking partial profits and moving your initial stop on the balance up to close to breakeven. For an idea of how much to move a stop, look at the average price rhythm for the secondary trend (the trend opposite your position).
Hopper Heuristic
At the hopper level?where prospective trades begin their journey?I am watching the major trend (directional movement):
? Price rhythm
? Time rhythm
? Volatility
? Thickness
? GSCS formations?Return, Double Intersection, Goodman Wave
These provide me a basic feel for the market in question. Should it become a candidate, I already have a good understanding of what is happening. At the hopper level, whatever your particular methods, you are only wanting to have a general understanding or feel for a particular market.
Candidate Heuristic
Once a prospective pair is cleared to be in the hopper, I watch for more specific signs to move it to candidate status.
A currency pair or cross cannot become a candidate on the basis of trend, price rhythm, time rhythm, volatility, and/or thickness. It is the appearance of a potential GSCS formation that moves it from hopper to candidate status.
At the candidate level I am mostly interested in ordinal GSCS formations.
If one of these occurs, or looks like it may occur, I am then watching for an entry point and calculating cardinal GSCS values. I am sometimes happy to trade without making extensive cardinal calculations. I have seen so many charts that I usually have a sense of what is fish and what is fowl.
?Ordinal? refers to direction only, with no specific values of price attached. ?Cardinal? refers to measurements with specific price values.
Trade Heuristic
I keep both a daily and a weekly trade plan. This is much like a doctor?s log for a patient. I can pick it up anytime I ?see the patient? and know the situation, ready to absorb new information. This was an idea I picked up from Frank Semone, and it is an excellent tool?clean, simple, and effective.
I get a buy or sell signal from a GSCS formation. Since I have been monitoring things from hopper to candidate, I am already prepared to move.
Money Management Heuristic
I set stop losses and price objectives based on the cardinal GSCS measurements and market environment (ME) elements of time rhythm and price rhythm. I may trade with no stop loss entered in the market?although I strongly recommend against the new trader doing this?but I never leave a session without both stop-loss and take-profit prices in the market.
I recommend the new trader use fixed dollar stops and objectives, placing them in as soon as a trade is executed?and leave them alone. If it is a long-term trade, you may wish to gradually move up your stops to protect
a portion of your profit.
Attitude Heuristic
I do not open a trading session if I am not 100 percent emotionally ready to trade or if I have less than three hours of time. I prefer in-depth trading sessions of 8 to 10 hours, and you will often find me trading from 2:00 P.M. EST to midnight. If I have other things of importance to do, I complete those tasks so my mind is clear to trade. I close the door to my home office. My wife and son know that means to stay away. Sometimes I play classical music in the background. I keep a notebook ready to jot down important notes, reminders, things of interest, and memory joggers. I punch in and out of a trading session on a sheet?using an old time clock from my Hawaii futures office. I know how much time a week I have devoted to trading and how long the trading sessions were. Each month I like to calculate my profit/loss in pips per hour of trading. I keep very brief notes at the beginning of the session and end of the session noting my overall attitude.
Postmortem Heuristic
I keep accurate records of each trade?date, pair, time in, price in, time out, price out, lot size. I also note the formation that triggered entry and the basic ending values?win/loss, amount. All trading platforms keep this information and many of them allow you to pull it off into a spreadsheet. I prefer doing it by hand.
At the end of every 10 trades I review all my notes, looking for sources of both strength and weakness. I am especially interested in finding ME Profiles (see Chapter 28) in which I have done exceptionally well (or poorly). I know from experience that small changes to my method can make big changes to the bottom line. Eliminating one bad trade can move a campaign from the minus column to the plus column. I try to make small, evolutionary changes rather than large, revolutionary changes. I treat trading as a process and think carefully before I tinker or otherwise interrupt the process. I then move the data to a set of spreadsheets where, if necessary, the numbers can be more intensively crunched.
Chart Scales
I use three scaled bar charts for each market I follow. This may vary from time to time?and I may use a fourth or fifth chart scale for a confirming look at some specific in the heuristic. If I see a chart with a potential GSCS formation I will use it.
Primary Scales
? 5-minute chart?timing
? 1-hour chart?watch
? 1-day chart?trend
Secondary Scales
? 15-minute chart?watch
? 30-minute chart?watch
? 3-hour chart?trend
? 1-week chart?trend
Which ones I use vary to some degree, but I am always thinking in terms of a watch chart (15-minute, 30-minute), a major trend chart (3-hour, 1-day), and a timing chart (5-minute). I avoid watching the 5-minute chart, because it moves too much and makes small price jiggles look overly important. Few traders use the 1-day or weekly chart, but they are worth a look?not just for the major trend but to see how truly long-term many FOREX prices moves are. We can get excited about a 100-pip move but fail to remember the major trends can be several hundred pips in length and months in duration.
A KIS (KEEP IT SIMPLE) SYSTEM
Before moving on to my trades, let me show you a very simple system for trading. It is, I think, ideal for the new trader. You can use it later as a candidate heuristic when you have developed a more sophisticated trade heuristic.
?The trend is your friend? is an old saying in the market. It is perhaps the best of many market proverbs. It is very difficult to get badly hurt if you simply trade with the major trend. The rub?how do you determine the major trend?
Trend is relative. A trend on a 1-day chart may be composed of numerous smaller trends and trading or congestion areas. But it can generally be accepted that the bigger the chart scale on which a trend shows, the more reliable it is. Currencies do tend to run in very long-term trends (a tidbit of information well worth considering). If you do not believe me, take 10 minutes to examine some daily and weekly currency charts. The major trend
is also a function of your trader profile. A 15-minute chart might be major to a Guerrilla, while it is probably a timing chart for a Position Trader.
A trend, of course, may at any time stop, go into a congestion or sideways mode, or reverse completely. Again, the longer-term the trend, the more reliable it is. Nonetheless, one wants to get on a trend earlier rather than later. Predicting prices requires some degree of anticipation.
The simple system uses only two chart scales from those listed earlier, but without reference to the GSCS formations. There are six steps:
1. Look for a long-term trend on either the 1-day or 1-hour chart. Ideally, look for a long-term trend on the 1-day chart and a current subtrend in the same direction on the 1-hour chart.
2. Look for the price rhythm on the long-term trend for both the primary trend direction and the secondary or correction trend direction. If these rhythms have a close average, you have a candidate.
3. Wait for a price rhythm average move in the secondary trend opposite the primary trend.
4. Move to the 5-minute or 15-minute chart. Average the primary and secondary trends.
5. Enter the market on the first move in the direction of the primary trend after an average move of the secondary trend. I call this the Dagger entry principle.
6. Place your stop at approximately 50 percent of the value of the secondary trend above or below your entry price. Place your take-profit price at approximately 75 percent of the value of the primary trend above or below your entry price.
If you wish to try for a larger piece of the price pie: For every 10 or 20 pips the market moves in your direction, move your stop half that distance closer. You may also use the Dagger to raise or lower stops: Wait for the market to pull back toward your stop and then move back up or down. When it does this, raise or lower your stop to just below or above the pullback price. Figures 8.1, 8.2, and 8.3 are examples of the Dagger formation.
While the structure remains the same, the specifics of each market may make them look somewhat different. Every market is unique!
