The similarity system | Page 575 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Similarity: A Practical Approach 859 replies

Similarity Trading - The ZZBB Method 215 replies

Rainbow similarity 256 replies

Similarity Stuff 17 replies

Directional Similarity 0 replies

Attachments: The similarity system
Exit Attachments

The similarity system

  • #11,481
  • Nov 3, 2014 7:32am Nov 3, 2014 7:32am
Quoting Eurusdd
{quote} The probability of 93.75% is for n=2!!! The theorem stated that clearly!!! Good work though
Hi Eurusdd,

Thank you for your answer.
Do you use only n=2 ?
When I read the ultimate truth thread, I thought that we need to work with long sub-sequences.
But if we need only n=2, that makes calculation easier. Thats good.

Thank you,


Quoting jackprobe
{quote} Hi Jimsterk, Thanks for the code, but as i tried it, i got err message: "Error: object 'data5' not found". How can i put the bars data from MT4 / NT to buffer "data5" ? If i do it manually, i think it will time consume .... {quote} Thanks in advance. Edit: Does any one have used this (MT4 - R connection) ? http://www.forexfactory.com/showthread.php?t=260422 Or any othersimple example of using NT-R connection is also ok, thanks
Hi jackprobe,

I have exported the histrocal data from my mt4 into csv file.
Then the csv data is read like "data5 <- read.csv(-------)".
I use "R for mt4" made by 7bit to connect mt4 and R.
I do not know other interface...
my knowledge about programming is very poor...

One thing I want to note is that R is good to calculate vectors, but very bad to iterate.
For example, I have posted this picture few months ago:

Attached Image


The data is 1-hour eurusd and sample size is about 20000.
It took about 24 hours for iteration. very slow.
C++ or matlab or other softwares are faster than R.
So if you are not already familiar to R, I do not recommend to use R.
(I am R user because I am a time series analyst and R already has packages like arma, garch, copula, etc.)

Thank you,
  • #11,482
  • Nov 3, 2014 7:55am Nov 3, 2014 7:55am
Quoting FX-Jay
please don't buy yourself a drink after each one & later accuse me of making you an alcoholic. Best, Jay {image}
  • #11,483
  • Nov 3, 2014 7:58am Nov 3, 2014 7:58am
Quoting jimsterk
{quote} Hi Eurusdd, Thank you for your answer. Do you use only n=2 ? When I read the ultimate truth thread, I thought that we need to work with long sub-sequences. But if we need only n=2, that makes calculation easier. Thats good. Thank you, {quote} Hi jackprobe, I have exported the histrocal data from my mt4 into csv file. Then the csv data is read like "data5 <- read.csv(-------)". I use "R for mt4" made by 7bit to connect mt4 and R. I do not know other interface... my knowledge about programming is very poor... One thing I want to note is that...
Wow wow wow, thank you very much mate. I think i really got it now . I will re-think again about using R, may be we can use it for simple calculation without complex drawing? About C++, could you please give me a link or a clue, where can i get similar solution using it, instead of R ? ( i mean, for statistical calculation in which we usually use matlab or R ).

Thanks a lot mate.s
Similarity can change your life !
  • #11,484
  • Nov 3, 2014 8:07am Nov 3, 2014 8:07am
Quoting FX-Jay
For those that still didn't get it for some reason, here's a live example from today on EURUSD: 1- Whoever thinks H is a secret is delusional and is living in fantasy land. I don't think the value of a correct H could be any clearer. It has been discussed here thousands of times for tens of different systems. My H value for this example is 30 because on this particular chart (currency + TF) you need a minimum H=30 to satisfy the condition of 97% RZ vs. 3% TZ. 2- Extend the previously formed TZ with light blue Horizontal lines. 3- Wait for this area...
Thanks for the really clear explanation. That was very clear that now anyone could understand.

Just wondering how you are handling the strong up trends like below?

After the blue circle is broken, whereabouts do you look for an entry? Is there any specific patterns you look for?

Regards
J
Attached Image (click to enlarge)
Click to Enlarge Name: Capture.PNG Size: 39 KB
  • #11,485
  • Nov 3, 2014 8:17am Nov 3, 2014 8:17am
Quoting FX-Jay
For those that still didn't get it for some reason, here's a live example from today on EURUSD: 1- Whoever thinks H is a secret is delusional and is living in fantasy land. I don't think the value of a correct H could be any clearer. It has been discussed here thousands of times for tens of different systems. My H value for this example is 30 because on this particular chart (currency + TF) you need a minimum H=30 to satisfy the condition of 97% RZ vs. 3% TZ. 2- Extend the previously formed TZ with light blue Horizontal lines. 3- Wait for this area...

Dear FX-JAY

I see you to set a good example, because I saw it on other occasions.
Look at my chart EURJPY:

→ When you have a sell signal (with my own system) I can go find the marked zone?
→ Is my interpretation correct?

Thank you.
Natalia.
Attached Image (click to enlarge)
Click to Enlarge Name: Sin título.png Size: 29 KB
  • #11,486
  • Nov 3, 2014 8:58am Nov 3, 2014 8:58am
Quoting jlee425
{quote} Thanks for the really clear explanation. That was very clear that now anyone could understand. Just wondering how you are handling the strong up trends like below? After the blue circle is broken, whereabouts do you look for an entry? Is there any specific patterns you look for? Regards J {image}
Hi My Friend, if you 'learn' carefully your chart history. that 'mid candle zones' is the strongest point of Transient Zone Indicator from our friend 'Free Fox'. because that 'mid zones' is very strong confirmation for 'trend continuation' in all levels of TF. especially if that 'mid zones' appear more than one in a row..

So if you found it in higher TF and lower TF support it, then go head ..

hope it help and best regards.
MTH
  • #11,487
  • Edited 9:22am Nov 3, 2014 9:01am | Edited 9:22am
Quoting FX-Jay
For those that still didn't get it for some reason, here's a live example from today on EURUSD: 1- Whoever thinks H is a secret is delusional and is living in fantasy land. I don't think the value of a correct H could be any clearer. It has been discussed here thousands of times for tens of different systems. My H value for this example is 30 because on this particular chart (currency + TF) you need a minimum H=30 to satisfy the condition of 97% RZ vs. 3% TZ. 2- Extend the previously formed TZ with light blue Horizontal lines. 3- Wait for this area...
Jay thanks for the clear Chart, the only thing still not clear for me is the value of "H"...whats the right H-value for 1Min TF / USD/EUR? and why is that so clear? i do not understand you guys always have to speak in secret's..?! if the value of H is that clear why you don't just tell us the value for it? sorry don't want to argue, i'm thankful for your Posts (the only clear to me) but just my thoughts..cheers mex
  • #11,488
  • Nov 3, 2014 9:32am Nov 3, 2014 9:32am
Quoting mex2000
{quote} Jay thanks for the clear Chart, the only thing still not clear for me is the value of "H"...whats the right H-value for 1Min TF / USD/EUR? and why is that so clear? i do not understand you guys always have to speak in secret's..?! if the value of H is that clear why you don't just tell us the value for it? sorry don't want to argue, i'm thankful for your Posts (the only clear to me) but just my thoughts..cheers mex
How to find the right value for h has been posted many times in this thread and in the TZ basics thread from vlady... And today FX-Jay just posted it again.
When h is correct you should have 97% recurrent prices (or bars) and only 3% true transient prices (bars).
How to do that? Look for the indicators from kprsa and freefox.

Nothing cryptic or secret just read again...
the 3rd wave is the crowd's wave
  • #11,489
  • Nov 3, 2014 9:38am Nov 3, 2014 9:38am
Quoting mex2000
{quote} Jay thanks for the clear Chart, the only thing still not clear for me is the value of "H"...whats the right H-value for 1Min TF / USD/EUR? and why is that so clear? i do not understand you guys always have to speak in secret's..?! if the value of H is that clear why you don't just tell us the value for it? sorry don't want to argue, i'm thankful for your Posts (the only clear to me) but just my thoughts..cheers mex
There is no secret if you exert effort and read.

Sometimes things are deliberately made secret. Because we don't want to spoon feed you.
Some people worked hard to discover how to make this system work. And it's not to be given away as if it is free or that everyone has a "right" to obtain it.
The secrets are given to people who demonstrate that they deserve it.

"If you have to ask, you'll never know" - Red Hot Chili Peppers
A pip is only worth it if you know how much you risked to earn it
  • #11,490
  • Nov 3, 2014 9:40am Nov 3, 2014 9:40am
Quoting juriFX
{quote} How to find the right value for h has been posted many times in this thread and in the TZ basics thread from vlady... And today FX-Jay just posted it again. When h is correct you should have 97% recurrent prices (or bars) and only 3% true transient prices (bars). How to do that? Look for the indicators from kprsa and freefox. Nothing cryptic or secret just read again...
No wonder 95% of retail traders are on loosing side

most of us (recurrent 97%) unable to read and understand things that repeatedly presented. Let alone the ability to tweak/apply to their trading strategy (me included...haha)
an attempt of compounding, from USD50+25bonus until ....burnout..haha
  • #11,491
  • Edited 10:35am Nov 3, 2014 10:15am | Edited 10:35am
Quoting jimsterk
Hi Kiads and Zelo, You two go too fast, I have a lot of stuff to catch up. One thing I am worried is that the sample data(=n^2) is large and we need to work with (n+1) increasing or decreasing sequences. I think n=2 is not large enough. (I am not sure though) If we use n=3, we have cal <- function(tri, da){ trials = tri res <- numeric(4) for (j in 3:6){ # 3-open, 4-high, 5-low, 6-close success = 0 for (i in 1:trials){ data <- da x = data[(ii+9)),j] if ( ((x[1]<x[2] && x[2]<x[3] && x[3]<x[4]) || (x[1]>x[2] && x[2]>x[3] && x[3]>x[4])) || ((x[2]<x[3]...
Hi jimsterk, thanks for posting your code. It has helped me debug my code. I have worked via a slightly different algorithm a comparison with your numbers. My data isn't the same as your data so there will be differences. Here is what I got for EU M5 (this compares to your post quoted above, the top example).

seqlen 10
subseq 3
minstep 1
maxstep 3
O, H, L C
1] 0.8162633 0.8345391 0.8383046 0.8162633

And for comparison this is with rnorm data:
[1] 0.6395818 0.6408301 0.6371818 0.6374646

So you can see that there is an approximate ~19% edge in the OHLC data vs rnorm data. In other words, OHLC isn't very close to rnorm when it comes to this particular sequencing method. Now, is it possible to extract this advantage, and if so, how? These are questions I would like to pursue with you guys who are already contributing / showing interest in this area with either R or MT4 or something else: jimsterk, stt, Zelo, Kiads, jackprobe and maybe some others. I would be willing to share my R code with you guys who want to pursue this. Just pm me and we can discuss how to implement this so as to capture the advantage. The code is in R and has been generalized to handle (presumably) any type of sequence.

Also by comparison, this matches my interpretation of Eurusdd's proposition with my EU M5 data and then compared with rnorm:
seqlen 5
subseq 2
minstep 1
maxstep 2
O, H, L C
[1] 0.8674744 0.8725777 0.8767430 0.8674744

rnorm:
[1] 0.8340689 0.8347517 0.8326345 0.8338138

So as you can see, the implied advantage is much larger with the first example, vs this example when compared with normal data. And in all cases the Low price had the highest probability. This is testing 145K EU M5 bars.
  • #11,492
  • Nov 3, 2014 10:19am Nov 3, 2014 10:19am
Quoting Kiads
{quote} LOL... I believe this is look like you... {image} and while you still trying hard to make sure yourself about that 'k' .. and stubborn enough with h=8 equal to 4 left and 4 right bars...lol. I'm testing the great theorem of chief EURUSDD.. and yes indeed make real money from that.. and real chartist should understand when look at this picture here where the money came from.. when probability higher then 97% and where that recurrent and transient zones actually are... {image} keep learning dude.. you deserved it to compensate your...
Hi Kiads,

regarding to your chart, you get these signals at the bar open? Thank you!

Regards
  • #11,493
  • Nov 3, 2014 10:24am Nov 3, 2014 10:24am
Quoting killerno
{quote} Hi Kiads, regarding to your chart, you get these signals at the bar open? Thank you! Regards
it's according to sub sequences theorem base on Zelo indicator with n=2 and i develop further for 'testing' purpose. That 'bomb' sign show up when conditions met and when current price higher from previous candle low, or lower than previous candle high. and will create continues horizontal line until 'touch' by future price movement.

best regards
MTH
  • #11,494
  • Nov 3, 2014 10:27am Nov 3, 2014 10:27am
Quoting Kiads
{quote} it's according to sub sequences theorem base on Zelo indicator with n=2 and i develop further for 'testing' purpose. That 'bomb' sign show up when conditions met and when current price higher from previous candle low, or lower than previous candle high. and will create continues horizontal line until 'touch' by future price movement. best regards MTH
Thanks a lot for the clarification Kiads!!

Regards
  • #11,495
  • Nov 3, 2014 11:00am Nov 3, 2014 11:00am
Quoting FXEZ
{quote} Hi jimsterk, thanks for posting your code. It has helped me debug my code. I have worked via a slightly different algorithm a comparison with your numbers. My data isn't the same as your data so there will be differences. Here is what I got for EU M5 (this compares to your post quoted above, the top example). seqlen 10 subseq 3 minstep 1 maxstep 3 O, H, L C 1] 0.8162633 0.8345391 0.8383046 0.8162633 And for comparison this is with rnorm data: [1] 0.6395818 0.6408301 0.6371818 0.6374646 So you can see that there is an approximate ~19% edge...


OHLC cross is the real 'median' of zones.. OHLC cross of 1 daily candle, actually are 'median' of 1440 1 M candles series.. that represent zone/area of price movement...

Attached Image (click to enlarge)
Click to Enlarge Name: Higher Probability of OHLC.jpg Size: 239 KB


all this years 'they' hide this, and not make it as the 7th option of MA price field.. lol.

MTH
  • #11,496
  • Nov 3, 2014 12:09pm Nov 3, 2014 12:09pm
Quoting FX-Jay
In the below Daily chart, there are 11 trade opportunities using the concept presented. ALL of them are winners & H is as small as 10. Once you are able to identify all 11, you can consider yourself as a graduate. Edit: In some of the 11 opportunities there could be more than one profitable trade done. However, I'm only talking about the main 11 opportunities (don't include sub-opportunities) Best, Jay {image}
Hi FX-Jay,

Do you have stats for the hit rate when K=0 for the above?

I'm getting about 85% for K=0 and H=10 in GU Daily. About 15% of the time, the price didn't come back to the zone.
Focus, Patience, Determination & Order in chaos
  • #11,497
  • Nov 3, 2014 1:09pm Nov 3, 2014 1:09pm
May I ask that the indicator, Transient Zones, be recoded with a buffer value for k=0. In return I will provide an iCustom() indicator that will give optimal k entry.
candy is dandy but liquor is quicker
  • #11,498
  • Nov 3, 2014 1:10pm Nov 3, 2014 1:10pm
Quoting saver0
{quote} Hi FX-Jay, Do you have stats for the hit rate when K=0 for the above? I'm getting about 85% for K=0 and H=10 in GU Daily. About 15% of the time, the price didn't come back to the zone.
In theory for H=10 on GU Daily Chart the probability of having a TZ is only 9% (so a 91% chance of RZ).
The above %'s are from K's Indicator, using a sample of 4,400 bars (i.e for the period ranging from Dec 1997 till today)

In order to get the hit rate you are asking for, we need an indi/expert that can calculate the number of 2 consecutive TZ. When we have this info we can derive the % we are looking for.

In theory, the % hit rate with K=0 should be >91%.
There's one probabilistic theory that could give you a value of 99%.
99.19% = 100% - (9%*9%)
However, I personally believe that is flawed, check the below link to understand better what I am talking about:
http://en.wikipedia.org/wiki/Gambler%27s_fallacy

85% doesn't look bad at all, but I believe the % should be >91%.

Whoever manages to create/code the above mentioned indi, please do share it with the rest.

Best,
Jay
  • #11,499
  • Edited 1:35pm Nov 3, 2014 1:24pm | Edited 1:35pm
Quoting FX-Jay
{quote} In theory for H=10 on GU Daily Chart the probability of having a TZ is only 9% (so a 91% chance of RZ).
The 9% chance of TZ is when it was being compared to all the number of bars which it looked at. It wasn't comparing it to the number of Recurrent zones. I think this is what most may not have gotten (or what I don't get).

For a zone to be recurrent, it has to be H-Transient on the previous number of H bars (left side), correct?
Then the number of RZ bars won't be equal to Number_Of_Bars_In_Chart - TZ_Bars.
We would have to actually find RZs that showed up exactly at the previous TZ levels and this number is very small.

Thats why there is 85% probability. It is however possible to get > 90% by tweaking K as EURUSDD mentioned, but then the number of trades per day drops as K & H increases.
Focus, Patience, Determination & Order in chaos
  • #11,500
  • Nov 3, 2014 1:36pm Nov 3, 2014 1:36pm
Quoting FX-Jay
{quote} In theory for H=10 on GU Daily Chart the probability of having a TZ is only 9% (so a 91% chance of RZ). The above %'s are from K's Indicator, using a sample of 4,400 bars (i.e for the period ranging from Dec 1997 till today) In order to get the hit rate you are asking for, we need an indi/expert that can calculate the number of 2 consecutive TZ. When we have this info we can derive the % we are looking for. In theory, the % hit rate with K=0 should be >91%. There's one probabilistic theory that could give you a value of 99%. 99.19% = 100%...
I have those numbers and they will surprise you, double tops and triple tops have a role here, it's not so rare to have another tz (for the same price) after having the first one. (Double tops i.e)
8 traders viewing now