Authors note: While the article presented uses the Trading Systems Analysis Group's "Triangles" package, it is by no means required to use the techniques presented here. For decades, triangles have been observed visually, and the majority of traders continue to do so. The triangle package only helps to make the detection, filtering, and visualization easier and less time consuming than doing the same work by hand, but these techniques can be used by anyone without the package as well.

One of the largest problems with triangles is that most experts don't agree on the rules that define valid triangle patterns. Some experts require a certain type of triangle in an uptrend, while others say that the same type of triangle should only be used in a downtrend. Some say to take either side of a triangle breakout, while others say to only take a certain direction based upon some class of rules. The fact of the matter is that there are many ways to trade triangles successfully, as well as many ways to loose money with them.

What is going to be shown is not a guaranteed success to your trading. After all, this is trading and there are no guarantees at all. But you will see a method to make your triangle trades have a higher percentage of success.

The basic method of trading a triangle pattern is to wait for one to develop, and then take a long position if it breaks above the triangle's resistance or a short position if it breaks below the triangle's support. The general premise is that since a triangle is a consolidation pattern, then whichever way prices break out of the triangle should be the direction that they should continue, and you as a trader would then hop on board this breakout to acquire the profits.

The losses occur when the breakout from the triangle doesn't result in an up or down trend but instead continues to move sideways, whipsawing you in and out of the trade while you're trying to jump on the trend.

This technique is designed to reduce (not eliminate) the chances of that happening. It can be associated with the classical physics concept of "a body in motion tends to stay in motion, and a body at rest tends to stay at rest." (While classical physics has no true relevance here, it does get the point across as to the nature of the technique.) If you are watching a market that is trading sideways, there is a higher probability that any triangle pattern that develops is not a true consolidation pattern, but is just a common movement in an ordinary market. Therefore you don't have a real breakout, instead you just have steady sideways movement (more designed for ordinary support and resistance techniques, as opposed to a consolidation type).

The basis of this technique is to trade triangles that are in a trending state. This way, prices are actually consolidating from something. If prices are trending in a particular direction (showing control by the bulls/bears) and a triangle occurs, then we see that at the point of consolidation there is a struggle going on. If looking at a triangle in an uptrend and prices break to the upside then the triangle was a continuation pattern. If it breaks to the downside, then the triangle becomes a reversal pattern. Either way, it will typically show a "winner" at the point of the breakout. In a non-trending sideways channel, a triangle occurs due to normal fluctuations as opposed to a struggle for control.

While the chances of a price breakout from this type of pattern turning into a sideways market is still highly possible, we have removed a very significant percentage of the failing patterns that would typically occur from taking all available triangles. "Qualifying" the triangle with a known parameter (a trend) increases the odds of success.

There are several ways to identify a trend condition. It can be observed visually (which is the primary way it has been done through time), or using today's software we can use indicators to determine if a trend exists. The indicator approach will be used for the examples here.

Many people choose to make their own trending indicator, and there are no problems with that. Since what defines a trend is fairly subjective, you should always use the indicator that fits best into your trading philosophy and style. But there are also several fairly good trend-identifying indicators that are already built-in to most software programs. Indicators like Wilder's ADX, r-Squared, or the Vertical-Horizontal Filter (VHF) are all valid indicators that can be used.

When using a trend-identifying indicator, the trend length will want to be matched up with an appropriate triangle size. For example, r-Squared and the Vertical-Horizontal Filter are both fairly fast indicators and therefore you would want to use them with fairly small triangle patterns. If you use an r-Squared indicator with a large triangle, then you run into the probability that the r-squared is identifying a short-term trend within the triangle instead of the trend that is causing the triangle to occur. A 21-period r-Squared will typically work well with triangles that are 8-bars or less. More than that will cause it to either not show a trend when the triangle occurs, or to show the trend within the triangle.

For the examples here, we will use the 14-period Average Directional Movement Index (ADX). This was chosen because it has a slower response than either the r-Squared or the Vertical-Horizontal Filter, and can therefore allow us to use larger triangles.

A value of 30 was chosen to be the threshold for determining whether a trend existed or not. If a value of 45 - 60 were chosen it would be a greater qualifier, but would also result in a less likely chance of detection. Since a triangle involves up and down movement to be created, by nature it will reduce the value of the indicator when the triangle occurs. Good triangles may force the indicator below the threshold values before it can be taken advantage of.

The application of this method is as simple as visually spotting a triangle pattern, and then applying the ADX indicator to the chart to see what its reading is. If the ADX is above 30, then the triangle is qualified.

Here is an example of how all of this works:

Here are actually several good triangle patterns. While the ADX was at its highest point at the beginning of January, there was a very small triangle (about 5 bars long at the beginning of January, or longer if you show the resistance back to late December) that resulted in a successful breakout just before the middle of January. Then towards the end of January another small triangle pattern developed, with the ADX still being above 30, as well as another successful trade to the upside that could have been taken. In mid February, another triangle setup occurs that looks almost exactly like the January triangle, except the ADX is below 30, so the trade would never have been made. At the end of the chart there is a very definite triangle. The ADX is above 30 and is starting to pull back (since the triangle pattern has significant down moves within it.) The trend is up (according to the +DI and -DI indicators) and so either a trend continuation or trend reversal would be expected.

The prices broke out to the short side with a potential gain of almost 20% in less than 2-weeks had a short position been taken. Also notice that at the end of the short downtrend that another small triangle develops, ultimately turning into a sideways market. The ADX is far below 30 at that point so this triangle would have never been taken.

Example 2:

Here, a triangle develops during a definite downtrend. The ADX is still around 40 even after an expected pullback. The breakout could be taken on either side.

What happened is something that occurs frequently with triangles. In this case, the triangle broke out to both sides during the next two consecutive days. It then huddled around those prices for several days before taking off to the upside in mid March. Often a triangle will have a false breakout only to reverse harshly in the opposite direction (as seen here). A trader with a good understanding of this type of phenomenon realizes that when they occur, it is best to be onboard for the sudden move. The small loss that occurred due to the false breakout would have been minimal to the potential profits by closing the position and taking the long side (almost another 20% in 3 weeks).

Example 3:

Here, a triangle develops with definite breakout levels established, as well as a strong ADX reading to the downside.

A breakout occurs on the eighth trading day into March, pulls back slightly, and then jumps to a short-term high in less than two weeks. Assuming a trader was willing to wait, then a large profit could have been acquired by waiting until mid-April for profit taking (though this type of position would have been difficult to maintain for that long without sufficient reason for staying in).

Example 4:

A much longer and larger triangle, though still within a defined uptrend as shown by the ADX reading above 30.

This is another common occurrence with triangle patterns that should be watched for. A breakout to the downside occurs with a complete pullback to the apex (middle line), and then forcefully moves in the original direction of the breakout. This is another example of how getting stopped out and reentering the position can result in a substantial return, more than offsetting the potential losses that can occur from getting stopped out initially.

Note that it is also possible that at the retracement to the apex, that the triangle could have been redrawn to extend to the middle of March. This would also be a valid triangle pattern, but would have failed the qualifying criteria since at that point the ADX was far below 30. (At the TSAGroup, we believe that it is better to be excluded from potential gain than to be exposed to substantial losses, so not taking the trade based upon that perspective would be consistent with our philosophy. You should determine ahead of time what perspective you will take on such a scenario.)

Final Notes:

Qualifying a pattern is nothing new, and the concept certainly shouldn't be excluded from visual patterns such as triangles. The ADX was used for its availability in most software, but remember that there are hundreds of trend-defining indicators that already do exist or can easily be created. You may want to use the ADX with more periods and at different thresholds for a stronger trend, or a completely new indicator based upon your own ideas. Visually defining a trend also works well, as trends can sometimes be seen even though the indicators don't show it. Find a method that works for you.

For Strategies Disk 2 - Triangles users:

For the examples here, we used the 14-period ADX to identify the trend, and we used the "TSAG - Triangles (Medium and Small)" exploration to find the ones we wanted. The original exploration filter was set to:

colA OR colB OR colC OR colD

We changed the filter to look for stocks that were priced above $10.00/share (on the American market) and with the proper ADX reading by using the following filter:

(colA OR colB OR colC OR colD) AND ADX(14)>30 AND C>10

Since the colA, colB, etc. use OR statements, we wanted to keep the precedence in order by enclosing that section of the statement in parenthesis. We then added our other criteria. This is the same way that you can add your own mechanical criteria to your explorations, which will also narrow the available signals that are returned each night.