by Darrell Jobman
As that subset of traders known as trend followers, we obviously are on the lookout for “trends”. Of course, how to define or measure a trend is itself subject to some interpretation, but for right now, let’s start with the types of indicators that us Turtles were taught to use. Generally, they can be classified in the category of “channel breakouts”. This is our weapon of choice because of its ease of use.
We start by defining a “channel” as some kind of horizontal or sideways price movement of a set of daily price bars. The price bars stay within a narrow range, with the highs never going above the top of the range and the lows never going below the bottom of the range. The range is the setup for the trade. What we’re looking for now is price action that penetrates out of that channel in one direction or the other.
As it is written for the computer program, we are just simply looking for “today’s price to be the highest high or the lowest low of the past X number of bars”. For example, in the specific model we use in our forex trading, “X” is set to 21 bars. In other words, we are going to go long every time the price goes higher than it has been for the past 21 bars and go short every time the price goes lower than it has been for the past 21 bars. This is our basic initiation or entry signal. It’s very simple and straightforward.
It is very important to keep in mind here that the exact value for X is not crucial to the systems; it is the concept itself that is much more important. The beauty of using a program like TradeStation is that you can test many different range variables simultaneously with the push of a few buttons. If the concept of trend following is valid and the concept of using range breakouts to try to measure or define the beginning or end of trends is also valid, then it should not matter much if we use 21 bars, 22 bars, or 19 bars to generate our basic signals.
In fact, if we would have found that altering the range length slightly wound up producing a large difference in total results, then we would have had a much more serious problem of statistical reliability. But if all of the various range lengths show consistently close results, that means our system is what the statisticians call “robust”.
Using this system, look at how easy it becomes to make trading decisions. Every day, my computer program (you can also do updates manually if you like), looks over all of the charts of the different markets that we might be trading, which takes maybe five minutes. We are searching for the markets that have been trading in a sideways consolidation for some period of time – the longer, the better, actually.
We then look at what price that market would have to reach to “break out” of the channel or range. For example, what price would be higher than the highs of the last 21 days, if that’s the specific rule we are using; or what price would be lower than the lows of the last 21 days.
We don’t know ahead of time what the direction of the breakout will be, but we want to be prepared either way so we don’t miss the move. So we set two orders in advance with our broker. If prices pass our price to the upside, we will enter the market going long. If they pass our price to the downside, we will enter the market going short. Technically, this is how we do it: We enter a buy stop order above the range and a sell stop order below the range.
Then we wait. If or when the market hits our price, we’re in. If it doesn’t happen, we simply repeat the same procedure the following day. Eventually, when the market moves far enough one way or the other, our position will be initiated.
When it comes to exiting trades, it is basically the same thing. If we are long, we will consider the uptrend to be over if/when the market turns around and goes lower for a certain number of days or bars. If we are short, the downtrend is considered violated if/when the market turns around and rallies higher for a certain number of days or bars. This is also simply written in computer language the same way as the entry rule above. For the forex trading model, the exit parameter (or “X”) is equal to 11 bars.
This strategy produced several highly profitable trend trades in the euro as shown in the chart below.
After an extended period of up and down price movement within a price range, we did not try to predict when or which way prices might move. Instead, as a Turtle Trader, when the price did exceed the high of the previous 21 days, we were prepared to go long on a buy stop. Then, after an uptrend, we exited when prices dropped below the low of the previous 11 days. We did not get in at the very bottom or out at the every top, but we took more than $6,000 profit from this trend.
The same basic strategy applies to going short as well: Sell when the euro drops below the low of the previous 21 days, cover to exit the position when the price moves above the high of the previous 11 days.
It isn’t just the euro or other currencies where the Turtle trend-following system works well. The same strategy produces similar returns in other markets as well.
Some items of interest should be noted:
- The exit signal is more sensitive than the entry signal. It takes more price movement to get us into the trade and less to get us out. Not every signal leads to a trend that extends enough to produce the sizable profits talked about here, but with a more sensitive exit signal, losses can be reduced if the extended trend does not materialize as anticipated.
- The exact values of the channel length don’t (shouldn’t) really matter. The truth is that Richard Dennis originally taught the Turtles to use 20 days to enter and 10 days to exit. This idea works well on computer testing (as does 21 and 11), but turned out not to perform as well in real life as more traders learned what values the Turtles used. As the number of people who anticipated what the Turtles would do at a specific price increase, they began “front-running” our stops in many of the markets, resulting in bad fills and increased slippage. So a couple of years ago, I decided to change the parameters slightly to 21 days in and 11 days out. The composite results of all of the computer testing are very similar, and I am sure the fills are better in real life.