Fractals, ZZ & the Pissing Dog | Page 13 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

ECN && STP && Scalping && Hedging - Broker Review - Oct 2009 24 replies

Alert on Fractals (MT) 2 replies

Zigzag & Fractals 10 replies

EA wanted for sound alert depending on fractals condition 2 replies

Fractals Edge Fx 3 replies

Attachments: Fractals, ZZ & the Pissing Dog
Exit Attachments

Fractals, ZZ & the Pissing Dog

  • #241
  • Jan 17, 2014 9:17am Jan 17, 2014 9:17am
Quoting candero
{quote} G, That's interesting, can I ask were those the only downsides you experienced? Apart from repainting and consecutive signals you didn't have no signal quality deterioration? I'm asking this because I've done it as a constant updating one myself. Thanks! C
candero,

I didn't check for a long time period so I can't compare results.
Those two side-effects were enough for me to abandon this route.

G.
  • #242
  • Jan 17, 2014 11:20am Jan 17, 2014 11:20am
GG53, just so we are on the same page, what sort of numbers do you have for your training data set, eg how many set1, set2, and set3 items do you have? For example using 5 min EURUSD from 2011 to 2012 (inclusive), I have 16368 items in SET 1, 12330 items in SET 2 and 1822 items in SET 3.

Thanks
  • #243
  • Jan 17, 2014 4:09pm Jan 17, 2014 4:09pm
Quoting beachloup
GG53, just so we are on the same page, what sort of numbers do you have for your training data set, eg how many set1, set2, and set3 items do you have? For example using 5 min EURUSD from 2011 to 2012 (inclusive), I have 16368 items in SET 1, 12330 items in SET 2 and 1822 items in SET 3. Thanks
beachloup,

2 years of 5 Min. data can't be just ~16000 items (SET1)...

G.
  • #244
  • Jan 17, 2014 5:02pm Jan 17, 2014 5:02pm
beachloup,

I have 85164 bars of 5-min data.
SET1 = 62406
SET3 = 850

Keep in mind I only have 295 trading days of 5-min bars (little over 59 weeks). So I don't have 2 years yet.

Carl
  • #245
  • Jan 17, 2014 9:42pm Jan 17, 2014 9:42pm
Hi GG,

could you please post a new screenshot from last week(any day). Would be useful to compare your signals to ours and analyse after.
Thanks a lot!

regards
  • #246
  • Jan 18, 2014 1:12am Jan 18, 2014 1:12am
Hi GG53,

I have some questions about step 3 of the algo, opposite bar prediction.

Question 1.
When i read the posts of the previous thread(thruth) I did read several times (see posts 888 and 902) that you start with the minimum length bar
in the range [-3,-2,-1,0,+1,+2,+3], I also read prediction for [previous bar, actual bar, next bar].
Carl's document says [actual bar, next bar] and you reviewed his document.
So i'm a little confused, Can you help me to get out of this.

Question 2.
I'm in a bearisch leg and in step 3 I'm looking for a 1 (opposite) bar prediction.
For instance actual bar prediction is 1, next bar prediction is 1.
But actual bar is 0 in reality.
Is this a nogo, means waiting for next bar.

Best regards

Huub
It is not important how often you fall down, but how often you get up again
  • #247
  • Jan 18, 2014 1:27am Jan 18, 2014 1:27am
Quoting crodzilla
beachloup, I have 85164 bars of 5-min data. SET1 = 62406 SET3 = 850 Keep in mind I only have 295 trading days of 5-min bars (little over 59 weeks). So I don't have 2 years yet. Carl
Hi Carl,

Your SET1 is based on patterns from length 4 up to and including length 13.
Or do you go above 13 ?

Best regards

Huub
It is not important how often you fall down, but how often you get up again
  • #248
  • Jan 18, 2014 2:06am Jan 18, 2014 2:06am
Quoting beachloup
GG53, just so we are on the same page, what sort of numbers do you have for your training data set, eg how many set1, set2, and set3 items do you have? For example using 5 min EURUSD from 2011 to 2012 (inclusive), I have 16368 items in SET 1, 12330 items in SET 2 and 1822 items in SET 3. Thanks
Hi,

Now it is getting interesting.
5M TF is ~75.000 bars a year
So 2 years should be ~150.000 records.
Substrings from 4-13, all possible combinations is (16 + 32 ..... + 8192) is 16368, so SET1 should be 16368 records independent of how many bars raw data.
If not enough raw data some records/pattern wil have no occurrences. Then you have some records with no occurrences in SET1.

SET2 gives about 4000 individual legs (every leg is a record) a year, so your 12230 is a little high, looks more to three years.
My SET2 is 7863.

For SET3 I have 1962 records, which is comparable to yours.

When i read GG53 and Carls reactions at least there must be some misunderstanding or talking about different pair of shoes.
Some clarification is needed.

BR.

Huub
It is not important how often you fall down, but how often you get up again
  • #249
  • Jan 18, 2014 4:20am Jan 18, 2014 4:20am
Quoting emonts
{quote} Hi, Now it is getting interesting. 5M TF is ~75.000 bars a year So 2 years should be ~150.000 records. Substrings from 4-13, all possible combinations is (16 + 32 ..... + 8192) is 16368, so SET1 should be 16368 records independent of how many bars raw data. If not enough raw data some records/pattern wil have no occurrences. Then you have some records with no occurrences in SET1. SET2 gives about 4000 individual legs (every leg is a record) a year, so your 12230 is a little high, looks more to three years. My SET2 is 7863. For SET3 I have...
I'm glad I asked the question now. I think my SET1 should be accurate because as emonts shows there is only a fixed number of combinations available when building substrings 4-13. If the numbers are wrong then I have made a mistake in my code (possible) or not understood the concept to create SET1 (likely).

To create SET1 I followed the method in post 22, I have put a screenshot of my C# code that is fully commented so if anybody can spot the error then please let me know. I can explain the differences in SET2 and SET3 but first I want to make sure I have the basics correct and try to understand why SET1 is wrong.

Attached Image (click to enlarge)
Click to Enlarge Name: SET1.png Size: 101 KB
  • #250
  • Jan 18, 2014 6:45am Jan 18, 2014 6:45am
Quoting emonts
Hi GG53, I have some questions about step 3 of the algo, opposite bar prediction. Question 1. When i read the posts of the previous thread(thruth) I did read several times (see posts 888 and 902) that you start with the minimum length bar in the range [-3,-2,-1,0,+1,+2,+3], I also read prediction for [previous bar, actual bar, next bar]. Carl's document says [actual bar, next bar] and you reviewed his document. So i'm a little confused, Can you help me to get out of this. Question 2. I'm in a bearisch leg and in step 3 I'm looking for a 1 (opposite)...
emonts,

Previous thread is a mess...
Here we start from fresh - with all documents in one place - so ignore previous thread and try to start fresh.
If you have doubts about something in the Doc's - ask the author of that doc and maybe we need to update it.

G.
  • #251
  • Jan 18, 2014 6:51am Jan 18, 2014 6:51am
Quoting emonts
{quote} Hi, Now it is getting interesting. 5M TF is ~75.000 bars a year So 2 years should be ~150.000 records. Substrings from 4-13, all possible combinations is (16 + 32 ..... + 8192) is 16368, so SET1 should be 16368 records independent of how many bars raw data. If not enough raw data some records/pattern wil have no occurrences. Then you have some records with no occurrences in SET1. SET2 gives about 4000 individual legs (every leg is a record) a year, so your 12230 is a little high, looks more to three years. My SET2 is 7863. For SET3 I have...
emonts,

Indeed.
SET1 should contain every bar in that 2 years, not just the "strings" or "sequences" which are used for filtering and creating of SET3.
Maybe it's a different way of storage (record), but I don't know how it was implemented on others database.

G.
  • #252
  • Jan 18, 2014 7:20am Jan 18, 2014 7:20am
I don't know if this will be of any use, but I coded a script which writes all of the combinations from 0 through to 2222222222222 to an array, before writing them to a file called: 'All pattern combinations.dat', which can be found at: ......\experts\files, in your MT4 folder.

You can open the file using Notepad.

I used the convention of: "0" for the root node, "1" for lower and "2" for higher.
Attached File(s)
File Type: ex4 ODAT BINARY TREE ALGO.ex4   3 KB | 577 downloads
  • #253
  • Jan 18, 2014 7:48am Jan 18, 2014 7:48am
Quoting emonts
{quote} Hi Carl, Your SET1 is based on patterns from length 4 up to and including length 13. Or do you go above 13 ? Best regards Huub
My apologies.. I just checked my code.
I use 4-15 in my database, but I only use 4-13 to filter for SET3.

Carl
  • #254
  • Jan 18, 2014 7:53am Jan 18, 2014 7:53am
By changing my code to lengths 4-13, I get 16366 for my SET1

Carl
  • #255
  • Jan 18, 2014 8:08am Jan 18, 2014 8:08am
Quoting crodzilla
By changing my code to lengths 4-13, I get 16366 for my SET1 Carl
crodzilla,

Ok, problem/issue solved...
It's different implementation of same thing - with same results.
My SET1 is actual raw data - as is. The algo build the strings while comparing to SET2.
I just added a counter to my code - and it's 16,368.

If I caused the misunderstanding/mess - beers to everyone (again) are on me...

G.
  • #256
  • Jan 18, 2014 8:13am Jan 18, 2014 8:13am
  •  FerruFx
  • Joined May 2007 | Status: MT4/MT5 EAs/Indicators/Alerts coder | 6,542 Posts | Online Now
Quoting gg53
{quote}If I caused the misunderstanding/mess - beers to everyone (again) are on me... G.
Beers should be on us ... for all you are doing here with your "team"
MT4/MT5 EAs/Indicators/Alerts coder
  • #257
  • Jan 18, 2014 8:13am Jan 18, 2014 8:13am
Quoting gg53
{quote} emonts, Indeed. SET1 should contain every bar in that 2 years, not just the "strings" or "sequences" which are used for filtering and creating of SET3. Maybe it's a different way of storage (record), but I don't know how it was implemented on others database. G.
Hi,

I have the raw data in a different table, and of course a SET can consist of more tables.
And because you need the raw data for generating SET2 it is very logical that the raw data are part of SET1.
So all clear.


BR.

Huub
It is not important how often you fall down, but how often you get up again
  • #258
  • Jan 18, 2014 8:14am Jan 18, 2014 8:14am
Quoting FerruFx
{quote} Beers should be on us ... for all you are doing here with your "team"
Then we should need to know place and time
It is not important how often you fall down, but how often you get up again
  • #259
  • Jan 18, 2014 8:17am Jan 18, 2014 8:17am
Quoting One day!
I don't know if this will be of any use, but I coded a script which writes all of the combinations from 0 through to 2222222222222 to an array, before writing them to a file called: 'All pattern combinations.dat', which can be found at: ......expertsfiles, in your MT4 folder. You can open the file using Notepad. I used the convention of: "0" for the root node, "1" for lower and "2" for higher. {file}
One day!,

Thanks!
Added to post #1.

G.
  • #260
  • Jan 18, 2014 8:21am Jan 18, 2014 8:21am
Hi,

my Set's for 2011-2012:
SET1: 16368
SET2: 7977
SET3: 2114

Carl,
Coud you please review your probability doc file. The total length calculation is different in your doc compare how GG is calculating. (http://www.forexfactory.com/showthread.php?p=7206574#post7206574 ). Maybe we should go through the pdf and ask GG to confirm the steps.

GG
If you have time could you please post a newer 5min TF screenshot.

Thanks for all!

regards
0 traders viewing now