Topic: Candle patterns difficult

Hi guys!

I'm new to forex . I began to learn about strategies and look for the "experts" in forex in Brazil , so I could copy them. Most are old men who have 20 years of market or more . They say that only use classical technical analysis. So I downloaded the software and am trying to recreate those classic strategies , but most of them need to interpret patterns in grafic. I can not find indicators to do so. So I came up with a question : such a wonderful software was created without these features in mind. Analyze classic technique is outdated ?

Example: I tryed to use Candle Color indicator and Doji indicator to test a strategy that could locate these entry points , but can not verify all requirements that the strategy needs.

Re: Candle patterns difficult


It's almost discretionary thing you're trying to establish - for instance gap up and down, how do you define it. Or "small bullish/bearish candle", what is small? Or downtrend... Visually easy to see, but for algo trading there needs to be certainty and clear cut conditions.
Probably those patterns need custom indicator approach.

Re: Candle patterns difficult

I'm sure the author of this setup doesn't provide a backtest. I'm also sure you cannot find it ready in any software.

ATR must be lower than Level.

Re: Candle patterns difficult

Thank you for the quick replies . It looks like I need to learn more about the tool.

Re: Candle patterns difficult

Popov can you attached your strategy? Thanks

Re: Candle patterns difficult

As Footon noted the pattern is not clearly define and especially the part with the down trend.
We need a custom indicator if we want to detect the pattern more precisely (points 2 to 6).

Re: Candle patterns difficult

Friends, you did not include features to easily find these figures in the charts because it's a difficult thing to define and do, or because these techniques are not worth the work ? A lot I found on the internet about forex focuses on these figures.

I wonder if they are worth the work, so I can continue using the FSB to make strategies that do not need them , and I train my employees to test and operate manually with these figures.

Re: Candle patterns difficult

While the interpretation is pretty open ended on some of the candle patterns, the open source library ta-lib has the morningstar candle pattern as well as a ton of others.  I've called out to it as an external dll from a c# ctrader indicator before.  It wouldn't work in this context obviously but the code might give someone ideas for future indicators.

Re: Candle patterns difficult

Friends, you did not include features to easily find these figures in the charts because it's a difficult thing to define and do, or because these techniques are not worth the work ?

Our goal from day one of this project is to provide the most precise backtesting algorithm. We have never promised we will include all patterns in our software.

FSB Pro and EA Studio have everything a trader needs to receive balance like this with minimum efforts.

Do you think that the complex patterns can make a better result?

Re: Candle patterns difficult

Popov, i am studying all your videos . In one you show 23 % return in one week . The largest Brazilian guru achieve between 5 and 7% per month . They say anything over 5 % is almost impossible. Perhaps these poor results are due to use complex patterns ?

FSBPRO is so wonderful that I'm afraid to believe these incredible results .
I just want 10 % per month.

Re: Candle patterns difficult

Your guy is right. I intentionally used larger entry amount in order to make the result more significant. 

Actually the FSB Pro user don't care the shape of the entry or exit pattern. The only important thing is the shape of the balance chart.

Here is another example:

We do our strategies in the opposite way

Usually the newer users have their ideas for trading systems or fancy new indicators. They come to FSB Pro and want to implement their ideas. Many time it ends with a disappointment because either the system doesn't make profit or is too difficult to implement these 'modern' patterns.

Contrary to that, we have a completely different approach. We set criteria and run the Generator. We don't care what is the entry or the exit setup when the balance curve is smooth and rising.

It is also possible to run the Generator without criteria and later to sort and filter the collected strategies.

12 (edited by hannahis 2016-07-31 09:20:41)

Re: Candle patterns difficult

amenobh wrote:

Friends, you did not include features to easily find these figures in the charts because it's a difficult thing to define and do, or because these techniques are not worth the work ? A lot I found on the internet about forex focuses on these figures.

Hi Amenobh,

I think you need to understand the strengths and limitations of FSB to know how to use it to it's fullest and how to avoid pitfalls.

FSB is more of a rule based trading software, not a Pattern Recognition software.
You may be barking at the wrong tree.

You need to understand the features of each indicators and create the appropriate rules/opening/closing conditions to "instruct" your EA to carry out your trading plan/strategies.

From what I gathered from your sharing. 

1. You can spot a repeating patterns such as "Morning Star" etc but you have not quite figured out ways to translate all these pattern "recognition" into trading rules.

2. You are hoping that FSB have a pre-make trading rules in the list of existing indicators section that you can use, to help you spot these formation/patterns, unfortunately FSB doesn't pre-make it for you (unless you use the Generator and input these indicators and instruct the Generator to search for patterns using these indicators with certain parameters's another learning curve.

In the past, I mentioned that we need to input our own trading rules into the Generator/optimiser to "guide" them towards certain level of success in finding robust EA.  Effectively, what I was suggesting is that, as we input these trading rules (which is actually trading patterns), we are guiding the Generator to find profitable combinations that will work well with these trading patterns/rules.

3. FSB is make to search for profit...equity curve, not search for trading patterns.

FSB search for EA that fit certain criteria and profitability.  Hence you may get so call profitable EA that happen to fit the curve out of random probability in the past (historical data) but may not profitable in the future.

4. Not all backtesting systems are built the same.

Searching for profitable EA based on pattern recognition, in my opinion, is more robust cos patterns tend to repeats by itself and hence has higher success rate of future "predictiveness" since these patterns repeats by itself and hence it is able to reproduce higher consistent success in the future (again depending how robust and well thought out, are your trading rules). 

FSB backtesting rules are not able to "see" only see numbers and generate/optimise EA based on profitability and other criteria.

If you are looking for a software that is able to recognise pattern and search for EA based on pattern are looking at the wrong direction.  FSB is not the answer.

Pattern recognition software such as neural network are made to search of repeating patterns and find trading rules based on these patterns.  I'm currently testing out this software call Neuroshell.

With Neuroshell, it's very easy to create a EA within minutes if you know what indicators you want (without any programming skills needed).  It searches out for patterns and find the best patterns out of a range of patterns you input into the system.  It has many pre-existing candle patterns (although I'm not a fan of such patterns, it's too "vague" in my limited knowledge).  FSB is not the only software out there, whereby you can create EA within minutes.  Neuroshell has 800 indicators to use and the good part is, if your indicator is not there, I'm able to create my own complex indicators based on the existing features that allow me to build my own formula (if you know how to utilise the features).  I'm not here to promote this software cos I'm still don't know whether those patterns that it is able to identify, is really profitable in real/live trading...we sometimes get fantastic results but failed miserable in live/demo testing.  Hence, I have yet to fully verify Neuroshell's capability.

5. All software claim to be good in certain things...bottom line is, can it delivery the results?

At the end of the day, we are hoping the investment we put into our software would yield us good our search for profitable EA.  You would by now know whether "What you see, is what you get" in your experience with FSB, is it really that easy to churn out profitable EA?  Are those EA that are profitable (in FSB accurate backtest results), really are that accurate/profitable in real/demo trading?  I can tell that you are a discerning user.

Likewise, I'm still in the process of testing out Neuroshell, it claim it's able to identify patterns and hence now I'm training my neural network to identify some the trading patterns I'm using and hopefully, it can fine tune it better.  I've yet to test out the result of the EA created by Neuroshell, hence despite it's claim and ability...I can't say whether again.."what I see, is what I get".

Bottom line, if you really want to use those candle patterns in FSB, you really need the discipline to translate these patterns into trading rules for FSB to execute it on your behalf.

I started out FSB with my own sets of patterns and I have to, through many trials and errors put these trading rules together into a winning combinations....that's the discipline a lot of FSB traders lack, we like to search for easy fix solutions.

13 (edited by amenobh 2016-08-01 05:06:04)

Re: Candle patterns difficult

Thank you for sharing your knowledge!

Currently I changed my focus and do not want to know more indicators or researchs. I just want the generated strategies go through all tests. If they pass the tests in the FSB , I put them in demo account . If they passed the demo account I put them in real accounts . I will have two teams of robots , one on the reserve , and another in the field. When a robot playing badly , we replace . Hope this works , I'm testing this method .

I imagine that only with these tests and proper management of money will be enough to achieve success.

14 (edited by hannahis 2016-08-01 14:28:45)

Re: Candle patterns difficult

Hi Amenobh

There are many ways to use FSB software.

Generally, there are two groups of users,

1.  Use their trading skills/knowledge to input their trading theories and use FSB's capability to fine tune it.
2.  Another rely on the statistical capability (with or without trading knowledge) to find the optimal combination.

Sounds like you are using the Generator to find your winning formula.  Let's us know how is your progress.  I do believe with our own input of good trading theory, the Generator is capable in finding winning formula.  Just that I'm now too busy with my own trading development that I need to focus and do not have time to start a new project to maximise the full potential of the Generator. 

I'm doing similar to your approach, I have 3 groups of EA.  1. The testing group, 2. Shortlisted EA and the 3. Live trading group.

The Testing group are newly developed EA and after 2 wks to 2 month of testing, those good ones will go to the second group, Shortlisted EA and after 1 month to 2month of testing (depending how "stable" is the EA), those good ones (with PF above 4) will go to the Live Trading group.

Those new EA in Live account allocated with 0.01 lot first (just to test out how their respond to live signals). Each week, I look at the Live account results and see how each EA perform (compare their net profit, profit factor, win%).  Those good EA with good performance, I'll increase the lot size from 0.01 to 1 lot (depending how good are their performance).  Those not good EA, I'll I delete them away from Live account and either put them back to shortlisted group for further observations or improvement (modify some trading rules etc) or completely discard them (banish to my archive file).

Re: Candle patterns difficult

Very good. This method is more perfectionist than mine. I will share my progress soon.
Good to know that already works for you smile)

Re: Candle patterns difficult

hannahis wrote:

  Those not good EA, I'll I delete them away from Live account and either put them back to shortlisted group for further observations or improvement (modify some trading rules etc) or completely discard them (banish to my archive file).

Hannahis, how much time you wait in live account to decide a strategy is not so good and delete? Till live account almost 3 stages passing strategies in your method.. backtesting in fsb and your 1st and 2nd stages.. After these successfull and quite long trials, i wonder how fast you decide a strategy is needed to be deleted or moved back to previous stages. I wonder it is short period or you wait for stagnation period?

Also have another question for experienced all users FSB. In real live account how many strategies running simultaneously? Is there an upper limit for you? Because with FSB can obtain many qualified strategies but how much run simultaneously i am not sure. Actually i already have 5 good strategies (still going on testing) and number will increased when i have more time.. but question is when i have 50 strategies then how to trade with them? all in real account simultaneously? Can you tell your experiences with it please.

Re: Candle patterns difficult

Calculate the worst scenario of all robots working together and losing money , and see if your capital can hold out long enough for a reversal of this disaster.