The similarity system | Page 290 | 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

  • #5,781
  • Aug 16, 2014 8:01am Aug 16, 2014 8:01am
Quoting LITEchild
{quote} hey buddy, what resources are you using to learn mt4 programming?
Google search of Programming MT4 yielded this:
http://www.mtprogramming.com/mql4/
A pip is only worth it if you know how much you risked to earn it
  • #5,782
  • Aug 16, 2014 8:08am Aug 16, 2014 8:08am
Quoting vlady1974
{quote} Google search of Programming MT4 yielded this: http://www.mtprogramming.com/mql4/
  • #5,783
  • Aug 16, 2014 8:43am Aug 16, 2014 8:43am
Quoting vlady1974
Guys, Keep thinking! Write out the formulas and google terms. I don't have it yet, but take the lead off LiquidGenius. If you don't know something, u have to learn it and study it. So this morning, I've started learning how to program MT4. Sorry... No quick answers here. Philosophy breeds ideas and discussion. Answers are verifiable facts from observation... So we must learn to observe for ourselves after ideas like this have been presented. FreeFox, my mathematics is poor...so I rely heavily on linguistic logic. My interpretation is that even if...

Thanks for the reply vlady ... I understand what you are saying that sometimes you don't have to fully understand sometimes to benefit. But I do understand what Markovian states means. Markovoian means that bars are totally independent and they have NO consideration of the past. But therein lies the discrepancy which I highlighted in my previous post ...

http://www.forexfactory.com/showthread.php?p=7678820#post7678820

EURUSDD's similarity system is about identifying DS zones that are in the PAST, yet Markovoian processes have no reference to the past. Was merely asking EURUSDD for clarity on this matter. Would like to know how that is resolved? I think anyone who does a bit of research themselves on Markovoian processes (and understands) will see the apparent anomaly, Be good to know EURUSDD's thoughts on the matter to grasp what he his trying to impart ...
*** Shoot for the moon. Even if you miss, you'll land among the stars. ***
  • #5,784
  • Edited 9:26am Aug 16, 2014 8:47am | Edited 9:26am
Quoting Freefox
{quote} Thanks for the reply vlady ... I understand what you are saying that sometimes you don't have to fully understand sometimes to benefit. But I do understand what Markovian states means. Markovoian means that bars are totally independent and they have NO consideration of the past. But therein lies the discrepancy which I highlighted in my previous post ... http://www.forexfactory.com/showthread.php?p=7678820#post7678820 EURUSDD's similarity system is about identifying DS zones that are in the PAST, yet Markovoian processes have...
Freefox, what I am presenting here is not related to the similarity system. I am trying to open your eyes to the fact that even if you assume the market is random and bars are independent, you can beat the market and there are ways to calculate probabilities for hitting times.

So, this addition is to help you calculate which prices are likely to be re-visited within a specific period!

I said we should assume the systems is Markovian just so I can show you that assumption cannot prevent us from beating the market!!!

Let me make this point clear. All markets depend on the past. Don't mix theory with practice.
  • #5,785
  • Aug 16, 2014 8:49am Aug 16, 2014 8:49am
Quoting Eurusdd
Now if you get the h right for a given time frame and at the current price p, the previous bars never hit p, then probability is on your side for a hit within the next h bars because the probability that p is h-transient is very low. If your h is right, this probability should be about 3%. That is 97% of the time, at least one of the next h bars should hit p. This works in theory and in practice for any stochastic process similar to the ones that govern currency prices!!! Very important: If your h is correct, then the height of your rectangle should...
Got it!!!! Looks like I will be learning how to code, thanks so much!
  • #5,786
  • Aug 16, 2014 8:51am Aug 16, 2014 8:51am
Quoting Eurusdd
{quote} Freefox, what I am presenting here is not related to the similarity system. I am trying to open your eyes to the fact that even if you assume the market is random and bars are independent, you can beat the market and there are ways to calculate probabilities for hitting times. So, this addition is to help you calculate which prices are likely to be re-visited within a specific period! I said we should assume the systems is Markovian just so I can show you that assumption cannot prevent us from beating the market!!!


Ok, thanks EURUSDD .. understand what you are saying ... I assumed the lesson was related to Similarity System, but I get the distinction now
*** Shoot for the moon. Even if you miss, you'll land among the stars. ***
  • #5,787
  • Edited 9:24am Aug 16, 2014 8:54am | Edited 9:24am
To understand why Technical Analysis rules and you don't really need to know what drives the market, consider the following case.

Suppose I call you one morning and said "XXX the current price of eurusd is 1.3445. Will you buy or sell?"

Now, think about this for a moment? If you buy/sell without at least knowing how price got to the current state, you are gambling!!! Every financial guru, billionaire etc will never place a trade without looking at the history. Even fundamentalists do it all the time. This is because, the way price got to the current state matters and I will explain this soon.
  • #5,788
  • Aug 16, 2014 8:59am Aug 16, 2014 8:59am
Quoting Eurusdd
To understand why Technical Analysis rules and you don't really need to know what drives the market, consider the following case. Suppose I call you one morning and said "XXX the current price of eurusd is 1.3445. Will you buy or sell?" Now, think about this for a moment? If you buy/sell without at least knowing how price got to the current state, you are gambling!!! Every financial guru, billionaire etc will never place a trade without looking at the history. Even fundamentalist do it all the time. This is because, the way price got to the current...
Yep, in my own feeble way this is what I was trying to describe:

(A - Markovian) Whilst each event is independent of past (and of future)...

(B - Non-Markovian) ... our expectation of the event being observed (based on knowledge or awareness of past events) helps determine their future state.


p.s. Oh yes, this is not directly related to Similarity, although I can now begin to see the relationship to that as well. Check the 3-Plan thread (which I did this morning) and the pieces are there too. Anyways, back to MT4 programming! LOL

p.s.s. Thanks Eurusdd for setting the direction of the compass!
A pip is only worth it if you know how much you risked to earn it
  • #5,789
  • Aug 16, 2014 9:01am Aug 16, 2014 9:01am
Quoting Freefox
{quote} Ok, thanks EURUSDD .. understand what you are saying ... I assumed the lesson was related to Similarity System, but I get the distinction now
A pip is only worth it if you know how much you risked to earn it
  • #5,790
  • Aug 16, 2014 9:06am Aug 16, 2014 9:06am
It is very important you know how price got to its current state before you buy or sell. In other words, the HISTORY of price does matter.
There are three possibilities.

Consider A. Price was increasing before it got to A and the probability it will continue in that direction is greater than 50%. WHy?
Because top and bottoms like B occur with probability less than 1%, if you define a period which is greater than three bars. Check it yourself. Therefore, since the probability that B is a turning point is less that 1%, price is expected to continue to do what it was doing before it got to B.
Price was decreasing before it got to C and since the probability that C is a local maximum or local minimum is less that 1%, price is likely to continue in that direction.

Now, if you do not know what price was doing before it got to A or C, you are silly to buy or sell.
Attached Image (click to enlarge)
Click to Enlarge Name: Capture.PNG Size: 43 KB
  • #5,791
  • Aug 16, 2014 9:23am Aug 16, 2014 9:23am
In summary: Believe in technical analysis, not just any analysis but the right one! Believe that there is enough information in the history of price to give you an edge. When you are done, START looking for the right information. I believe I got you started!!!
  • #5,792
  • Aug 16, 2014 9:33am Aug 16, 2014 9:33am
This is getting very interesting! I've attached an indicator (public domain) which should be useful in playing around with different settings for 'h'. It's a fractal indicator where you can choose how many bars go to make up the fractal e.g. if you want 7 bars on either side, set the total = 15 and those on either side = 7. I have used these settings in the example on the M5 chart but this is in no way a recommendation, I haven't had time to play about with it yet to find the optimum.

In the example a few things come to light. The first rectangle (grey because it is subsequently taken out) isn't formed until the 7th bar after the fractal because we can't know it is a valid fractal until that point. By then another potential fractal is forming on the downside. When that fractal is confirmed, I believe we can then assume that all prices between the bottom of the previous rectangle and the top of the newly completed bottom rectangle are recurrent prices. So, with a known open rectangle above, we can open a buy with a significant degree of confidence that it will become profitable. And we don't have to hold the buy until the rectangle is reached, we can scalp away until we are happy with the profit or another intermediate rectangle is formed.

I suspect that 'k' is too large for the first red rectangle so I would not be relying on getting all the way there but would be happy to trade towards it from an upside rectangle in the knowledge that the prices in between should be recurrent.

The other point of interest is at A. The 7 bars to the left of A alert us to the fact that a fractal may be forming but at point B the price goes above A. So we can assume that point B is recurrent and a good place to trade towards this would be after the completion of the next upside rectangle. Sure enough, the price is revisited at C - by the time the upside fractal has fully formed, the price has already come down significantly but there are still 6 pips to play with so enough for a scalp.

Now to do some research on optimum settings.

Thanks again Eurusdd for your insights.
Attached Image (click to enlarge)
Click to Enlarge Name: EURUSDM5aug16v1.png Size: 24 KB
Attached File(s)
File Type: mq4 wlxFractals.mq4   5 KB | 1,109 downloads
  • #5,793
  • Aug 16, 2014 9:53am Aug 16, 2014 9:53am
Quoting Eurusdd
{image} Let me explain one way you can trade this fact. I borrowed this picture from a post above. Now at the current development, he placed a ? mark at the red spot because we do not know if the zone developing will be transient yet. But this is where you can beat the market. Let us suppose that A will be a transient zone, then we can expect that the region just above A must be recurrent. Therefore, once price gets into that region, we are in for $$. So, all you have to do is estimate a constant value for k or let is depend on the current bar....

If i got this right we're looking for transient zones. Once one bar brakes transient ZONE A then price should return and touch ZONE A to enter long position.
Attached Image
  • #5,794
  • Aug 16, 2014 9:55am Aug 16, 2014 9:55am
Quoting hepsibah
This is getting very interesting! I've attached an indicator (public domain) which should be useful in playing around with different settings for 'h'. It's a fractal indicator where you can choose how many bars go to make up the fractal e.g. if you want 7 bars on either side, set the total = 15 and those on either side = 7. I have used these settings in the example on the M5 chart but this is in no way a recommendation, I haven't had time to play about with it yet to find the optimum. In the example a few things come to light. The first rectangle...
Thank you so much, this is a great place to start. It works in all TF's?
  • #5,795
  • Aug 16, 2014 10:02am Aug 16, 2014 10:02am
Quoting hepsibah
This is getting very interesting! I've attached an indicator (public domain) which should be useful in playing around with different settings for 'h'. It's a fractal indicator where you can choose how many bars go to make up the fractal e.g. if you want 7 bars on either side, set the total = 15 and those on either side = 7. I have used these settings in the example on the M5 chart but this is in no way a recommendation, I haven't had time to play about with it yet to find the optimum. In the example a few things come to light. The first rectangle...
Interesting observations hepsibah, and a nice clear chart. But I don't think we can assume that price will return to point B. That price is recurrent because of point A. It doesn't mean it will be recurrent in the future too. Take the bar after the first red down arrow as an example. The bottom part of that bar is recurrent, yet the price never goes back there. Just my take on things anyway, I could be wrong!
You reap what you sow.
  • #5,796
  • Aug 16, 2014 10:08am Aug 16, 2014 10:08am
Also hepsibah, if h=7 then your indicator should have painted a fat rectangle halfway up bar B, shouldn't it? Price didn't go there 7 bars before and after.
You reap what you sow.
  • #5,797
  • Aug 16, 2014 10:10am Aug 16, 2014 10:10am
Quoting NorthTrader
{quote} Interesting observations hepsibah, and a nice clear chart. But I don't think we can assume that price will return to point B. That price is recurrent because of point A. It doesn't mean it will be recurrent in the future too. Take the bar after the first red down arrow as an example. The bottom part of that bar is recurrent, yet the price never goes back there. Just my take on things anyway, I could be wrong!
Yes you are wrong. That region was a possible recurrent region but because price never went there, there is nothing to trade. If price had gone below that red-region, then we can assume the red-region is transient and below it is recurrent. In that case, we expect prices below the red region to be re-visited almost surely.
  • #5,798
  • Aug 16, 2014 10:11am Aug 16, 2014 10:11am
Quoting NorthTrader
{quote} Interesting observations hepsibah, and a nice clear chart. But I don't think we can assume that price will return to point B. That price is recurrent because of point A. It doesn't mean it will be recurrent in the future too. Take the bar after the first red down arrow as an example. The bottom part of that bar is recurrent, yet the price never goes back there. Just my take on things anyway, I could be wrong!
He's just pointing out how the indicator works and an observation with his current settings, we would have to try to find the optimal settings for the TF of our choice. Still a lot of work to do. But a good place to start for the non coder.
  • #5,799
  • Aug 16, 2014 10:14am Aug 16, 2014 10:14am
Quoting davinci15
{quote} If i got this right we're looking for transient zones. Once one bar brakes transient ZONE A then price should return and touch ZONE A to enter long position. {image}
Everything before the bold is correct. You are not guaranteed longs. What you are guaranteed are re-visits. That is what you should trade!!! If you suspect a zone to be transient and price moves to a region which will then be recurrent, you can then trade the re-visits. But don't try to buy when price hits the top of the suspect zone because the zone is still forming!!!
  • #5,800
  • Aug 16, 2014 10:33am Aug 16, 2014 10:33am
Quoting Eurusdd
{quote} Everything before the bold is correct. You are not guaranteed longs. What you are guaranteed are re-visits. That is what you should trade!!! If you suspect a zone to be transient and price moves to a region which will then be recurrent, you can then trade the re-visits. But don't try to buy when price hits the top of the suspect zone because the zone is still forming!!!

Thank you for correction. One more question, does the price re-visit the transient zone within the h bars or could be latter?
2 traders viewing now