Forex Blog

First-hand Forex trading experience and information about foreign exchange market that will be useful to traders

Archives

Best Indicator for Divergence Trading?

March 3, 2014 by

Divergence between the price chart and the corresponding oscillator indicator levels is a well-known trading signal. It is based on the fact that if the momentum indicator (which should be non-laggging) fails to continue with the current trend direction, the trend’s exhaustion is to follow.

Here is an example of a bullish divergence signal. The price shows a new lower low, but the CCI indicator fails to show a lower low, signaling a probable reversal of the current bearish trend:

Bullish Divergence Example

And here is an opposite signal — a bearish divergence. It appears inside an uptrend when the price posts a new higher high, while CCI displays a lower high. Notably, this particular signal is a loser as can be clearly seen from the ensuing price action:

Bearish Divergence Example

Although the concept of divergence trading sounds very simple, there many nuances both in detection of the signals and in usage of these signals. You can refer to my own basic MACD strategy as a one example of what can be done with divergence.

Several technical indicators can be used to trade divergences: Moving Average Convergence/Divergence (MACD), Bill Williams’ Awesome Oscillator, Relative Strength Index (RSI), Commodity Channel Index (CCI), DeMarker, Stochastic Oscillator, etc. The main condition for the indicator to be capable of showing divergences is its non-lagging nature.

What is important to remember no matter which indicator you choose for divergence trading — not every divergence is a valid signal. Only sell on bearish divergence when it appears within an established uptrend, and buy on a bullish divergence only when it appears within a clear downtrend.

There are existing divergence indicators that help detect divergences automatically:

Disclaimer: these indicators were not developed by me.

Personally I find it difficult to spot “real” divergences with such automated indicators, whereas the signals themselves are also often unreliable. A test of RSI divergence trading performed by Thomas Bulkowski in stocks market showed rather poor results. Even though I believe it is possible to develop a winning methodology to trade divergence in Forex, there will always be too much left to trader’s own discretion to deter a proper automation of divergence strategies.

What indicator do you use for divergence trading?

View Results

Loading ... Loading ...

If you want to ask a question about divergence trading or if you want to share a detailed opinion on using various indicators for divergence trading, please feel free to use the form below.

15 Responses to “Best Indicator for Divergence Trading?”

  1. Sylvester

    Hello,
    I came across this site as I am interested in divergence trading. While researching different chart setups I came to conclusion that different indicators generate at different time valid divergence signals that could end up in profitable trades. Do you think that a working out a single divergence signal based on multiple technical indicators would make sense?
    I would appreciate you answer.
    Greetings,
    Sylvester

    Reply

    Andriy Moraru Reply:

    It is a very difficult task. It will also require a lot of backtesting to prove strategy’s viability.

    Reply

  2. Sylvester

    What if such task has already been completed? I developed a software that generates single regular and hidden divergence signals from 10 technical indicators plus a “turbo divergence” chart study that based on those signals generates formation prototypes, by drawing resistance or support lines. All the user needs is add a trigger study to such line to generate a buy or sell signal. I can check divergence between two points, in any range of bars or run it as a real time study. Same with the “turbo divergence”.
    Bottom line is that each of those indicators has its time when it is the only one to generate a divergence signal and it makes a perfect sense. When running on a 1 minute chart nothing can be hidden there, you see what happen on such chart from a beginning to the end. Line attributes include a list of indicators contributing to a given signal. It took me about 2 years to do the development (improvements!) and testing. Still have a lot of ideas to implement as — as it happens in such cases – once you open one door after a while you start to notice another.
    What do you think? Can you see any value here?

    Reply

    Andriy Moraru Reply:

    There is value if its trading signals can fare better than randomly generated signals. Did you try backtesting it?

    Reply

  3. Sylvester

    Are you refer to some kind of automated backtesting? As far as I know divergence trading cannot be automated therefore any kind of automation is not an option at all. Detection of divergence signals can be automated but not decisions about a trade. Too many things need to be considered. As it is based on formations, first you need to identify its beginning, that in some cases is not as straightforward. Sometimes divergence identify the whole formation, sometimes only the end of it. You may need to use a manual tool to find out the exact basic formation and its extensions. It is a pure art with very strict rules: when to buy, when to sell, when to abort a trade before reaching price target. However, from what I can see, while applying strictly those rules, with a proper mindset you will never be lost.

    Reply

    Andriy Moraru Reply:

    You need to have some objective scale to measure the usefulness of a tool. The simplest of such scale is the profit earned using the tool compared to the output generated by a random trading system. If you know other effective ways to gauge the usefulness of such an indicator, please explain.

    Reply

    Andriy Moraru Reply:

    You could look at this from the other point of view. If you believe that your method of finding divergences is superior to traditional ones, but the whole divergence trading cannot be handled by mechanical algorithm, you could just test for comparing your indicator with others. You could set up a backtest to test some very simple divergence trading system on your indicator and on the traditional ones. In that case, you are not necessarily expecting to see profit from your indicator, but it should show better results than those obtained with traditional divergence indicators.

    Reply

  4. Sylvester

    It looks like there is some miscommunication here…
    My system uses standard/traditional technical indicators, calculated in a standard way, that generates divergence signals same as your system above. The only difference is that if two or more indicators generates same signal it will be represented on a chart as a single one.
    I might see more signals as other indicators can view the chart in different way. I filter out all conflicting signals so the chart looks nice. As I said before, the most important is to identify formation. In your case using a single indicator in some specific situation your formation can appear too tall, you will calculate the price target, make a trade and the price will never reach the target. In the same chart scenario another indicator may create a divergence signal for a shorter formation, and trading that shorter one will succeed. I have seen plenty of situation like that one. In my opinion, this is the most important aspect in divergence trading. This is also the reason why you need the backtesting. In case of my system the output is fully consistent regardless of time span. If the consistency of my brain were the same, I would become a reach person in a short period of time. However, it looks like it was made up for engineering of software systems…
    I made a collection of technical indicators based on a real need. I saw a formation on a chart and none of my indicators signaled a divergence. So I had to find one. Now I have 10 technical indicators and almost no problems in identifying all important formations, however the most difficult is when a divergence signal appears at the bottom of formation. To identify such formation I need to find its beginning… I am planning to develop a manual study that will help me with that.

    In my opinion, taking into account charts complexity and all the mechanisms I have discovered so far, automated divergence trading does not make too much sense for me. What makes sense is to use such system as I have and have a team of people (at least two) each of them watching different aspects on a chart. E.g. Having two people on a team, each of them would monitor one direction and both of them would need to agree to a trade.
    You wrote: “You need to have some objective scale to measure the usefulness of a tool”
    In my case based on generated divergence signals you can break a chart into regular or hidden divergence formations, same of them are tradable, some two small to touch, but sometimes extremely important.
    Thanks to those divergence signals I was able to discover some very interesting tricks (someone has to enforce them!), it really helps to understand what happens on a chart.
    On Friday on a single chart I discovered 5 trading opportunities: 3 regular divergences and 2 hidden. All of them reached their targets.

    Reply

    Andriy Moraru Reply:

    That’s all nice, but without some strict objective rules of application of your divergence signals, you cannot know whether your trading results can be attributed to those divergence signals or to some way of discretionary thinking you have applied during the trading decision taking process.

    Reply

    Jim Reply:

    I trade with limited divergence and would like to talk with you.

    Reply

  5. Sylvester

    It looks like I have made a horrible mistake by coming to this site and giving away my email address. Spam, spam and more spam. Are you traders or spammers?

    Reply

    Andriy Moraru Reply:

    What? We never send out spam. Please forward any messages that you believe were unsolicitedly sent by us to webmaster@earnforex.com, so we could report their original sender.

    Reply

  6. Sylvester

    If you do not send spam then my email address was shared with spammers. As I did not shared my email account with anybody in months, this would be the only explanations of all that rubbish my mailbox was overloaded with today.
    Ad rem:
    Divergence signals come first, in particular hidden divergence and regular divergence (in the form of formation resistance/support lines). Then comes the strict rules, then human decision. It’s a complete system. In your case, you have automated system, rules are encoded into that system by the system designer or when you tune the system to its best performance: at that time you apply “your discretional thinking” :)
    Both systems have the same fundamental elements but the way those elements participate differs.

    Reply

  7. Sylvester

    Hey Jim!
    What’s up?

    Reply

  8. Bj

    Hey Sylvester I would like to talk to you

    Reply

Leave a Reply

required
required (will not be published)