Free Coding request

Please check, wether, this one gives the correct result.
(I have overlooked a +1 additon in one an algorythms :S )

If it works, I will modify the rectangles and for the text plcement tomorrow. (However i'm not sure, that the rectangles can be moved tot he background.)

Thanks Gorraf.
In the indicator you posted last, it fixed the recognition issue but now it counts 4 cases but only displays 3 cases. (that is all the available data)








One thing to consider.
If you want to consider a huge bar with a minimum pips of say 1000, and of length one suppose 'U' to see how a currency responds to major news/events,movements, the indicator would require me to place minimum pips to 1000. But, that means that in all likeliness any moves after that will likely be considered Neutral if the next bar isn't as large as 1000 pips.
As you can see, it makes it complicated. Perhaps you can make a separate input variable "Maximum pips to be considered neutral"/'Max pips Neutral" so the variables don't get too complex regarding the minimum pips to be considered variable.

Some good news, and some bad news.

The good news:

1. I have split the pip input into two separate inputs.
   Now you need to add a number, which defines the Neutral limit in the pattern and a number which defines the Neutral limit in the aftermath bars.
2. I have made a button, which makes the rectangles go into the background or calls them forth, depending on button state.

The bad news:

1. I was not able to reproduce the error, you have encountered with the weekly EUR/SEK chart.
   The indicator found five results in the same timeframe and even more on a longer period. Was there any other input vale set, that could help me narrow down, where hte issue might be?
2. The rectangles aren't perfect, and there isn't really anything that could be done about them.
   - If they are in the background, only their outline is showing which hides the wicks of the candles
   - If they are in the foreground, their border disappears, and the filling color will be visible.
   a) if you are inspecting only one result, or give them only a one letter pattern, then, since the rectangle is basically one dimensional, and the borders do not show, it will disappear from sight.
   b) if there is a correlation between two rectangles, their color might change (according to the rgb palette) on the correlated territory, or if they are the same color, they neutralize each other

Hello Gorraf, Thanks for the Updated version.

In terms of the problem I found, try openning the USD/SEK on the weekly charts and inputting the pattern UUUDUD. My broker is limmitted to May 2012 so an input up to there would be fine maybe 500 bars.
On the graph it shows 3 instances but the indicator states 4 cases, which ultimately messes with the probability. The image attached shows all available data to me from pepperstone upto sometime in 2012. So that was the problem.
Maybe it is because it is counting the new forming pattern and since the result would be neutral, the errors may be from that.

Ah okay, yeh when indicators overlap their colours mix and become something else.

I'm curious, are you finding this indicator useful?

Actually, what really bugs me, is the fact, that, while the candle at 01.17.2016. is Bearish on my screen, which has opened at 8.58240, and closed at 8.56370. On your screenshot, it appears to be a Bullish, and as of that, it didn't get registered.

The issue is the same with the candle from 06.12.2016, which is a Bearish again (Opened at: 8.30864, and closed at: 8.23066) on my screen, and a Bullish on yours.

I have made several typos (again) in the last post, including, stating, that I did not found the issue on the EUR/SEK chart, while on the screenshot, it is clearly visible that I meant USD/SEK, but nor that, nor the time I spent looking at your chart and at the code didn't brought me closer where the 4th match, or the bug could be.

The marking and the counting are working from the same pool of matches, so it is really strange, that the two does not match up, hovewer on your second screenshot from post Nr 22, I have found something, that needed to be fixed. (I also fixed the jumping textm, and button. It got pretty annoying, while I was seeking the problem.)


On a side note:
The bar history is a Metatrader setting, which can find under:
Head bar / Tools > Settings > Charts
if you want to see more.

Actually let's leave the USD/SEK, instead lets consider the EURUSD Weekly (If our EURUSD data doesn't match one of us has dodgy brokers).

Parameters
Tf: H4
U/D Pattern : UDUDUD
Sum of Inspected Bars After Pattern :1
Min pips Pattern :0
Min Pips Follow-up :0
Maximum Bar range :250


Since 8 matches are stated but only 6 appear on the graph, I hypothesized perhaps the two missing matches are not appearing because the follow-up bars were Neutral. So I changed the Min Pips follow-up to 1000 to see if any of the other existing patterns disappear from the display/graph. I realize that if it were neural it would simply appear as a red bar, but something strange happens in the lower example.

Parameters
Tf: H4
U/D Pattern : UDUDUD
Sum of Inspected Bars After Pattern :1
Min pips Pattern :0
Min Pips Follow-up :1000
Maximum Bar range :250


This time, only 6 cases are found and 6 appear.

Hello There are not always bugs, but sometimes there are. I found another one.
For some reason it reads as Up.

Hello Gorraf I hope you find the available time to work on this indicator.


In the mean time you may be interested in working on another one aswell. I'll refer to this as 'coverage'. Basically how much a candle engulfs it's previous candle.
Suppose H1 and H2 are highs(including wicks) of candles 1 and 2. respectively, L1 L2 lows of candles 1 and 2.

if H2> H1 then (replace) H2=H1
if L2<L1 then (replace) L2=L1
equate
(H2-L2)/(H1-L1)

apply to all sets of two consecutive candles, find mean specific to currency and timeframe. return withmean, median? standard deviation and finally a 95% confidence interval based on normal distribution or something similiar.

only if you are interested

I have added a switch, which should make the analyzing more accurate, but I am a bit clueless about what the actual bug might be.

Thanks for narrowing down the problem in post Nr. 26, this actually helps me (or at least it should), but since our charts do not match, I can not reproduce it, and test it a further. Could you tell me, at which broker you are, and what time zone it uses? Hopefully that will lead me somewhere.

In the post Nr. 27, I'm pretty sure, you have added, "-1" at the inspected bars, that's why a down reads as an up, because the result searching goes backwards.


Anyway, I fixed the texts, so hopefully, they will not jump back and forth at every refreshing.


I will look into your new request in the meantime.

I took a peek at your "coverage" indicator.

I have coded it, which you can find attached.

Also, while looking at it, I thought, that it would be visually more appealing, if the formula would be: 1-(H2-L2)/(H1-L1)
Due to this, where a candle is smaller than the candle before it, the hystogram will show a 0, and depending on how much a candle engulfs the previous candle, the bar grows. Which you can find also attached to this post.

Actually, looking at that I thought, about why not dig into your basic idea even deeper, and I have altered it even more. The hystogram bars in the third indicator are separated into two halves and the top part shows the positive, while the bottom part the negative engulfing.

This is some quality material.
Thanks for all of this.
I will check it all out. In post number 27 I didn't have it as -1 I just tested it again with strange returns...

Did the EUR/USD data matchup?

I'm with pepperstone and in metatrader the time is synced with gmt +2.

Okay the biggest bug I am experiencing is the counting problem in post number #26 where the cases found doesn't match the cases shown on the graph. This is evident when you get a probability less than 50. Like just now I tested EURUSD Daily for DUUU on the last 100 bars, it shows 7 matches but the cases found states 9. Hence the probability is lower. I don't know whether we have different data (lol) matters in this case. please try and test this out on your platform and look for when the indicator is returning the wrong number of cases. Perhaps it might be good to start again with the code, if the bug can't be found.
It's kind of disheartening at just how hard it is to find high probability set ups, I guess the more hard work and time you put in, the more likely it will be to find a good pattern set up. Thanks and goodluck.




Regarding the coverage indicator, just to confirm, the Highs and lows which are measured include wick?
Okay so this is sort of embarrassing but I used Paint to draw this example.
In the picture since only 1/3 of the 2nd candle covers the previous candle, recognise it as 1/3.
Since the 3rd candle covers everything , recognise it as 1. (H2-L2)/(H1-L1) (Assuming H2 AND L2 are bounded by H1 and L1)
I believe the difference here between your indicator is that here it is measuring how much the current candle engulfs the previous, where as in the indicator it is checking how much the previous candle engulfs the current.
Using"ProblemSolving_Hmsr_coverage_ind.ex4 " So the coverage would be less than 1 in both cases. It looks like (H1-L1)/(H2-L2) (Assuming Instead that H1 AND L1 are bounded by H2 and L2)

That was kind of the idea I was hoping towards. Maybe I'm reading the graph wrong. The aim was to see for any given candle, how much is the next candle going to overlap into the current indicator. Whereas I beleive currently the indicator shows how much the current candle overlaps into the next candle.

This is pretty useful. As you can see, the coverage suggests that the current weekly candle will likely cover atleast 30% of the next weekly candle if there are no shocks etc. But since we don't know the size of the next weeks candle, it is not useful really just yet. Where as if instead it measured how much the last candle covers the previous candle, and suppose the minimum over the last few months was also 30%, one could expect that each new candle will cover the previous candle by atleast 30% without shocks

PS ignore the other probability indicator

I was starting to write a leghty answer, but then I realized, there might be a viewpoint issue.

In charting, the actual bar, which is forming is marked as 0, the latest complete bar is marked with 1, the next to the left is 2 and so forth....

So, when, in your example, you refer to H1, L1 you refer to the first bar to the left, and refering H2, L2 as the second, or you count them, as H1, L1 as the first bar you inspecting, and H2, L2 the second from the left?

This struck me, when I was looking at your example and read "Since the 3rd candle covers everything".

No you are right in the H2 L2 is for the second candle, etc, and I meant the 3rd candle measures everything in the second candle etc.

I guess what I was tryng to say is
1>for candle 2 bound the high and low so that they are not larger than the highs and lows of Candle 1.
2>take the distance of the bounded second candle and divide it by the length of the first candle. then plot on histogram.

And I believe, what is currently on the indicator is
1>for candle 1 bound the high and low so that they are not larger than the highs and lows of Candle 2.
2>take the distance of the bounded first candle and divide it by the length of the second candle. then plot on histogram.

Regards

Your graphical explanation helped me a lot, understanding what you wanted, so there isn't really anything embarrassing about it.

At the first indicator, I simply went with the formula you wrote, and didn't thought much about the solution, but I have completely rewritten it now. Check it out, whether this is what you had in mind.
(Hopefully tomorrow I will have some time to revise the other indicator as well.)

I am new to this forum. I want to built a EA by from yours. If you help us ı will happy. Thank you very much.
I want to buy and sell only GER30(DAX). My language is very low but hopefully you can understand
My strategy, for example; for x second- y pips increase my account will sell,ıf opposite to this stiuation x second y pips decrease my account will sell. If my Account not buy or sell until my time, program will reset to all time and it will be start again.for example program kindly turn by turn.My aim, calculate to speed of part of exchange and catch to turn back. thank you very much for your help { email address deleted by staff }

Dear abbus!

Please open a separate thread, or send me a PM.
Hopefully, you understand, that I don't really would like to see our bughunt drailed.

Thanks.

TY Gorraf, Your Coverage Indicator is exactly what I was hoping for thanks alot. It's also great to see you coding a various range of meta trader tools. We all need to make the most of you while you are here

Dear gorraf, thank you very much for message but ı have no idea about open PM.Could you please give information about how can ı do it. I had writing mistake for my first message , this message is true

''
I am new to this forum. I want to built a EA by from yours. If you help us ı will happy. Thank you very much. I want to buy and sell only GER30(DAX). My language is very low but hopefully you can understand

My strategy, for example; for x second- y pips increase my account will sell,ıf opposite to this stiuation x second y pips decrease my account will buy. If my Account not buy or sell until my time, program will reset to all time and it will be start again.for example program kindly turn by turn.My aim, calculate to speed of part of...''''

I feel myself like I just found the pin in the haystack.

Making an account at pepperstone and being able to see what you did made the trick.