Pattern-Finding EA

Im gonna work with this a little bit over the weekend

I have been coding EAs for the past 5 months since the whole shut down and I've made about 50 of them. Saw this thread and decided to make an account to help out this thread. Going to read the code over today and get back to you guys!

can you please coding for fixed lot size because if you change lot size EA dosenot take command and used its own lot size.

One more thing its work well on fast moving pair like Gbp and its crosses and best result is After London session.

Yes I wondered what had happened. Cannot wait to try updated version

i will be following your progress on this.

1. I noticed in the code the logic when calculating the lot size it can be improved. RR didn't code properly but within we can code a try catch statement that will guarantee a proper lot size, I haven't debugged and made sure I am right but once Monday rolls over I can properly test

1. Thank you, I have cleaned the code up and made it more readable for now. Some basic warnings came with the original code but I fixed all 9 of them already. I split the original code up into its respective sections. Tomorrow I will try to re-code RR's functions to execute the same actions just more efficiently. Once I understand RR's logic perfectly I will start to improve this EA. Lets see where this goes!

I would appreciate anyone posting their results on different pairs, parameters, timeframes, timezone, etc!

Sounds good. I plan to run the EA on a demo account from next week. To understand how it operates better..

ive never experienced this. I set a lot size and it will execute that size. I am also just running this as a test. I trade futures so I would have to manually execute all trades taken by this EA(until(if) i get API set up with me broker). and it also needs to be coded in something like python or C++

If anyone is interested, here are the pattern finding indicator WmiFor and EA WmiFor30 mentioned in Post #35 by ForexTest.
The author is Murad Ismayilov.
I used Google Translate to get English versions.
Murad's explanation says WmiFor uses Spearman's coefficient for predictions.
WmiFor30 uses a faster, more efficient way to calculate (supposed to be more accurate).

There is a problem (array handling/size/reference) with WmiFor30 so it doesn't seem to compile correctly in latest MT4 version.
It is interesting to put WmiFor on the chart with PatternFindingCorrelation EA and compare predictions.

Haven't had a chance to do any forward testing.

Looking forward to SerbSerb's updated EA.

Stay safe,
NW

1. forexhard on the first page mentioned if this EA could learn from its self and like you said for that type of system we would need to add another layer in c++ or python to achieve it. I have one system in python that executed bitcoin trades, in the fucutre maybe we could morph the two together

1. Thank you for the post, I should have seen that comment and thought of this! Especially that I know a bit of Russian, if anything I will send this to some family if i really need to know exactly whats his post is saying. Sucks there isn't any comments at least in Russian but the code is very readable. I will fully document RR's EA first so I have a base of understanding then start debugging to understand how it reacts, see the data, error report, etc. Hopefully I can finish documenting today so tomorrow I can get right into debugging and move onto fixing Murad's indicator code.

pipguy mentioned adding a time filter which is super simple so I'll be adding that as well.

Yes I agree, it might be helpful to convert the original EA to a more robust lang like Python or C++

1. Thing about that is if we start researching correlation ratio indicators/EAs in other languages then this thread will never end and nothing will get done lol so lets focus on what we have here, perfect it then start discussing implementing the logic over.

Also Stanislav Korotky updated the Murad's Russian mt4 code into mt5 code here: https://www.mql5.com/en/code/27436. It has some fixes and bus as well apparently but ill take a look at it later.

Hi SerbSerb,

Here are the versions with original Russian comments.
Plus Murad's companion Volume prediction indicator.

Also, a lengthy, detailed explanation here:
https://forexsystemsru.com/threads/e...r30-dtw.69396/

1. Thank you for the code, gotta switch it all to Cyrillic but ill do that if the translated one doesn't make sense Reading the translated one from ForexTest its pretty bad so going through it now fixing it.

Hey, i want to learn how to develop trading algos in MQL4, but I'm finding it hard to learn, but i know a bit of python, Please do you know any resources i could use to learn to develop algos in python? And which one do you find easier?
Thanks.

LOL,
... if the translated one doesn't make sense...

When I pasted first Russian phrase into Google translate, Google recognized it as Vietnamese.
"Global Variables" was translated to "Always Pleasure".

Enjoying humor where & when we find it.
NW

1. Are you new to coding? I'm busy with this and other life stuff at the moment so i can help you once I know more about exactly what you want! For me researching and learning both MQL4/5 or python is relatively the same, just depends on how crazy you want to get into it lol python definitely has more capabilities plus is becoming more popular these days.

1. Yea the translating was not as simple as just throwing it into google translate LOL. I swapped from UTF-8 characters in the code → iso-8859-5 → Russian Cyrillic → English so here is a better description of the indicator settings:
   1. IsTopCorner: whether to show the indicator information block in the upper right corner. If false, it will be shown in the lower right corner.
      OffsetInBars: the offset of the sample. The default is 1 (the analysis starts with the first closed bar). This option allows you to shift the analysis to the past for exposures and checking the reliability of the indicator predictions, comparing its forecasts with actual price movements.
      IsOffsetStartFixed: is the start of the sample fixed? If yes, it can only be moved manually by mouse.
      IsOffsetEndFixed: is the end of the sample fixed? If yes, it can only be moved manually by mouse.
      PastInBars: The size of the sample in bars, which is searched on the history (from 3 and above). Please choose depending on the period of the chart. The longer the sample, the less history-like samples will be found. The default version is 3.0 - 12.
      VarShiftInBars: the maximum possible shift of the pattern on the X axis in bars. It only makes sense on timesheets below D1.
      ForecastInBars: How many bars forward make a forecast (from 1 and more). At your discretion. Affects only the visual perception. The default version is 3.0 - 4.
      MaxAgeInDays: on what period of history in days to look for similar patterns (by default for the last year, MaxAgeInDays = 365).
      MaxVarInPercents: Still need to translate lol
      MaxAlts: how many of the most similar samples are taken to build the corridor. The default is 5.
      ShowPastAverage: Whether to show the moving average on the sample.
      ShowForecastAverage: Whether to show a sliding average forecast.
      ShowBestPattern: Whether to show one of the possible price movements (automatically hides the moving average).
      ShowTip: Whether to show trade advice and SL/TP levels.
      IndicatorPastAverageColor: the color of the moving average sample.
      IndicatorCloudColor: the color of the projected moving average and cloud of options.
      IndicatorBestPatternColor: the color of one of the options.
      IndicatorVLinesColor: the color of vertical moving lines (beginning and end of sample).
      IndicatorTextColor: the color of the indicator's information block.
      IndicatorTakeProfitColor: the color of SL/TP lines.
      XCorner
      YCorner: coordinates of the indicator's information block relative to the upper right corner. If you use another indicator that also displays information, you can move the unit

   
   
   

1. That means I looked around the code and not sure if it was Murad's fault or someone else but their debugging by using log files literally will never get called cause their check statements before logging was always false. I decided to use this style of debugging instead (https://www.forexfactory.com/thread/...-for-debugging thank you 7bit ) and wow if only I googled this before lol. I found a little memory leak and toned down the data collecting to get it working at least. I didn't change any logic or anything else, just posting so over night hopefully people can break/test it! Not posting the source code cause I have it placed around with comments and other stuff I don't want to clean up before going to bed plus I don't plan on stopping With RR's EA I broke the code up, debugged a little and filled as many comments as I can could. Tomorrow if I get RR's EA done ill compare it the Murad's algorithm code, see which one is better.

thank you for the hard work Serb, looking forward to test the EA version.

for now im checking several timeframes with the indi you posted to see if theres any confluence or if indeed the projection can be reliable

Yes i am fairly new to coding. I started learning python about 3 months ago, and i completed the basics of the syntax in the "python crash course" book, but i have no knowledge of C++, which i hear MQL is based on, so i guess that's why i might be having problems understanding it properly. Thanks.