Keltner Channel Breakout Strategy

4
Keltner channels have been around as a charting technique since the 1960’s. Those traders that use them mostly do so to trade breakout events. They define these as events where the price makes a crossover of one of the Keltner lines.

Nonetheless some also view Keltner channels as a trend following tool because of the use of moving average lines.

Keltner channel breakout strategy
Keltner channel breakout strategy © forexop

The Indicator

There are several variations of the Keltner channel but the idea in most of these is the same. There’s a mid-line that forms the central axis. Then there’s an upper range line and a lower range line.

The upper and lower lines create the channel or band of the Keltner indicator. These measure deviations of the price around the central axis line much like the Bollinger band.

Main line = EMA (Period)
Upper line = Main line + S x ATR(Period),
Lower line = Main line – S x ATR(Period),

Here Period is the averaging width and S is the deviation size. A wider period will produce a smoother band. A large deviation will create a wider band. This is similar to the Bollinger band where the width is defined by the number of standard deviations from the center line. With both indicators the bands move wider apart as volatility increases and narrow as it decreases.

Figure 1: Keltner channels versus Bollinger bands
Figure 1: Keltner channels versus Bollinger bands © forexop

The Keltner channel tends to see more abrupt changes than the Bollinger band which is usually smoother. This is because the ATR, on which Keltner is based, doesn’t use simple averaging techniques.

How to Interpret the Keltner Channel

The classic way to trade the Keltner channel – the one suggested by Keltner himself in the 1960’s – is to enter long as the price breaks above the upper channel and to trade short when the price breaks below the lower channel. This is known as simple crossover and is a typical breakout strategy.

The table below shows three back tests covering a ten year time frame using this approach. The tests were done using the four hour chart (H4) with the following setup:

Keltner period: 50
Keltner SD: 4
Volume per trade: 0.01 lots
Maximum open: 5 positions

Symbol Profit ($) Trades Profit factor Expectancy (pips) Drawdown ($)
GBPUSD 310.92 1233 1.06 0.25 455.5
EURUSD -177.06 956 0.94 -0.19 481.41
USDJPY 142.02 1305 1.04 0.11 247.64

Limitations of the Crossover Method

As can be seen from the table, the returns are volatile. This is typical with breakout trading systems. The main reason is due to false breakouts or fake outs. These happen when the price breaks out of a channel temporarily, only to return to previous levels. If you have entered a position in the breakout direction, it will usually result in a loss.

The other consequence of this is that the drawdown is large as a percentage of the profit. In the GBPUSD test run the drawdown is 147% of the profit and for the USDJPY test this rise to 174%.

Keltner Bands Strategy

With the classic approach, the break of the channel obviously depends on how you set up the indicator. In the Keltner indicator, the width is defined by the deviation, which is just a multiple of the ATR (Average True Range).  The greater this multiple is the wider will be the channel.

There isn’t anything fundamentally significant about the width of the channel. For example there’s no reason why a break of a channel of width 2 should be significant but not a break of a channel of width 1.5 or 2.5 or 1.95. So why trade one and not the other?

Banding method

Instead of using a simple crossover approach as above, the strategy I prefer is to look at a whole range of points on the chart and see where the market strength lies. To do this I use two Keltner channels. Together these divide the chart into five bands.

As with classic Keltner this is purely a breakout strategy. I don’t use trend signals or any other input.

To use this method, you need to create a histogram that records the amount of time the price has been within each “band”. Obviously this is easier to do with software but if you prefer manual trading you can also do this quite easily by sight using a simple box counting method.

For simplicity I use two bands but there isn’t any reason why more can’t be used. The two bands divide the chart up into five zones. This is shown in Figure 2.

Figure 2: Keltner lines created from two channels divide the chart into 5 bands.
Figure 2: Keltner lines created from two channels divide the chart into 5 bands. © forexop

I then examine the chart over the lookback period. Each bar is placed into a bucket numbered H(0) to H(4) corresponding to the band in which the center of the bar lies. The tally of all of these bar positions then produces a histogram.

So for example, with a lookback period of 100 bars, the histogram might look something like this:

Bucket H(0) H(1) H(2) H(3) H(4)
Count 0 0 9 91 0

The count is just the number of times the center of a bar appeared in that band.

What does it tell us? A histogram is more useful than a single crossover event because it encapsulates far more information. In particular the shape of this histogram tells us where the market strength lies.

Skewed to the right: If the histogram is skewed to the right this suggests strength on the upside. The recent price action is tending towards the upper bands.

Skewed to the left: If the histogram is skewed to the left this suggests strength is on the downside. The recent price action is tending towards the lower bands.

Neutral: If the histogram is symmetric, this suggest the market is evenly balanced

It’s important to understand that this method doesn’t measure the market trend. The histogram can still be neutral even when the market is rising or falling strongly. Rather it measures “changes” that are developing where the market is diverging away from its recent path.

And because the histogram is calculated over multiple bars it doesn’t respond to single crossover events unless the price action moves upward or downward for a longer period. This reduces the chance of being caught by false breakouts.

Figure 3: Histogram shapes
Figure 3: Histogram shapes © forexop

Trading Decision

The trading rule is based entirely on the “5-band” histogram which is calculated at the current bar on the chart.

Buy entry signal (long): H(3)+H(4) > H(2)

Sell entry signal (short): H(0)+H(1) > H(2)

Stop losses/Take profits

The trade exits are set dynamically.

On the buy side, the long position is closed if the price breaches either the upper Keltner line or the lower Keltner line. This is the upper line of the widest Keltner channel. When the upper line is breached this will generally be a profitable trade and when the lower line is breached this will usually be a losing trade.

Figure 4: Example take profit and stop losses
Figure 4: Example take profit and stop losses © forexop

The sell side is the just the reverse of the above. The short position is closed if the price breaches the upper or lower of the outer Keltner lines.

I always put in a broker-side “hard stop loss” as well. This is just as a failsafe in case the dynamic stops fail to work for whatever reason.

Results

I back tested the system on the four hour chart (H4) over a ten year period. The volume was again one micro lot per trade (0.01) and a maximum of five open positions at any one time.

Keltner period: 50
Keltner SD outer band: 6
Keltner SD inner band: 3
Lookback period: 100 bars

Symbol Profit ($) Trades Profit factor Expectancy (pips) Drawdown ($)
GBPUSD 635.13 810 1.30 7.80 152.73
EURUSD 472.39 954 1.17 5.00 359.08
USDJPY 484.08 526 1.44 9.20 119.24

In all of these tests the profit is greater than maximum drawdown. This does suggest that the histogramming method reduces the volatility of returns by reducing entries into false breakouts.

Figures 5 and 6 below show the returns for this histogramming method. Figure 7 shows the classic Keltner strategy using crossover. The returns in this example are significantly more volatile than when histogramming is used.

The number and quality of trades can be adjusted by changing the width of the Keltner bands. Generally speaking, wider bands produce few trades but of higher confidence than do narrower bands.

Narrow bands can work well for high frequency, low profit trading – i.e scalping.

Figure 5: GBPUSD, four hour chart back test
Figure 5: GBPUSD, four hour chart back test © forexop
Figure 6: USDJPY, four hour chart back test
Figure 6: USDJPY, four hour chart back test © forexop
Figure 7: Classic Keltner shows much higher volatility of returns compared to banding method
Figure 7: Classic Keltner shows much higher volatility of returns compared to banding method © forexop

The Keltner channel indicator for Metatrader (MT4) can be downloaded here for free.

Download file
Please login

grid_ebook_thumb
How to implement a successful grid strategy: Step by step

This ebook is a must read for anyone using a grid trading strategy or who’s planning to do so. Grid trading is a powerful trading methodology but it’s full of traps for the unwary. This new edition includes brand new exclusive material and case studies with real examples.

A step by step approach to a successful grid strategy.

Join 11,000+ other traders and subscribe to Forexop's newsletter. It's free to join and you'll get updates directly to your inbox. Just add your email address below.
About the Author

Steve Connell has spent over 17 years working in the finance sector as a trader/market maker and strategist. Over that time he’s worked for several global banks and hedge funds. Steve has a unique insight into a range of financial markets from foreign exchange, commodities to options and futures.

4 Comments
  1. Ok, so you mean having two on the chart. One with an MA of 50 and one with an MA of 15 or so? I am fairly new to the game as you may have intuited!

    • It isn’t necessary to modify. You can do it all just using two copies of indicators on the same chart. One is set with the wider channel and one with the narrower channel. The bands are then defined by the mid line and the upper and lower bands of the two indicators.

Leave a Reply