The similarity system | Page 647 | 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

  • #12,921
  • Edited 5:08pm Nov 20, 2014 4:55pm | Edited 5:08pm
Quoting infinitus
{quote} Hi Tashkent, very valid question. When we dig through the literature of finance-mathematics, we will find so many different views and thesis, and we find, that people can hardly agree on anything. It is mainly because, out of my view, that one tries to frame market, which is basically driven by social mood, greed, fear, emotions, into a mathemathical formula. Every one who follows quant trader will notice, that their models work for some time and then stop producing the desired gains. What is the reason? We can go back only a few pages to...
i have read Hanover's post and i do share the same thoughts with him.
we all know that TS's ideas are really interesting and useful to some extend, no question. but i want to talk about his negative legacy.
his view of the market helped some people here to shape not so healthy mindset like "beating the market" - to be one step ahead of the market. they all think that there are levels, specific numbers which price must reach. if price does not go the way they projected, they call it a "fake move". they developed a habit of challenging the market in a disrespectful way, trying to make it to obey their calculations and obsession with closing every single trade for profit etc. such mindset never has led to a good ending.

that is why, i asked a few questions about their understanding of the market and their thought about the REAL REASONS behind the success of similarity or T/R zones. and the answer i was expecting was not just the statistical results. my question is much deeper than that.
As Above, So Below
  • #12,922
  • Nov 20, 2014 4:57pm Nov 20, 2014 4:57pm
  •  Neio
  • Joined Oct 2012 | Status: Can't think outside?Enlarge the box | 688 Posts
Quoting stt
{quote} just to correct this (as i many times corrected in EURUSDD's posts as well), in random market there is no predictability. you dont need to be math whiz to see this. His argument seems right when you have gamblers fallacy (i.e. if you think probability of Tail is higher after 10 Heads or as EURUSDD said, prob of one T in 7 toss higher after seeing 14 toss with all heads). In random market there can be no predictability - just think clearly and you will see why (hint: because random market means each price is independent of past moves).
Just to say that I was pointing out my understanding of what Eurusdd was suggesting when he introduced Markov to the thread.
An expert is a man who has made all the mistakes which can be made
  • #12,923
  • Nov 20, 2014 5:02pm Nov 20, 2014 5:02pm
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!!! Let...
Here is the post you all are tripping over.

Might I add; what a waste of time to keep on squabbling without reading the thread.
  • #12,924
  • Edited 5:24pm Nov 20, 2014 5:09pm | Edited 5:24pm
  •  Neio
  • Joined Oct 2012 | Status: Can't think outside?Enlarge the box | 688 Posts
Quoting infinitus
{quote} Haha, nice!!! I have questioned several times here, what the limitations of the system are. I questioned several times, why EURUSDD was not able to trade his findings successfully. I asked, in which situations we can gain money easily, and in which not. No one answered. That spoke for itself, I thought. Well, at least Kprsa noticed the problem in the past........ But I do not fully agree: it is not only the Markov property that fails (e.g. there seems to be market memory), it is also the probability rate which is way to high (regardless...
It was not a system as such but a way of looking at the market. Eurusdd did make a suggestion as to how it could be traded.

The thing is you say that the Markov process fails. Maybe it does but the indicators that have been produced based on the theorem as presented show that you can configure h so that 97% of prices are h-recurrent and only 3% are transient. The problem I see is that it is not a tradeable 97%...i.e most of the recurrence happens on the next bar or two and there is not enough gap between entry and the PTZ being cleared. I think someone produced some stats to back this up.
An expert is a man who has made all the mistakes which can be made
  • #12,925
  • Nov 20, 2014 5:14pm Nov 20, 2014 5:14pm
Quoting kprsa
{quote} Because of the self-similarity property of the price signal. For example, in order to have a change of trend, all the sub-cycles must complete (similarity). If they are not complete (dissimilarity) the trend is expected to still go on. In my opinion this is a sound technical view of the market. k

how do you define sub-cycle?

Is it simply cycle on lower TF?


Now we have to think for a while:


When we believe, that the lower TF do follow the higher TF (a typical trader meme: lower TF must follow higher TF, because higher TF rule etc., you know this stuff all, I think) we have a problem:

when we imply this, we would be living in a deterministic world. All events in the world would be needed to be designed in a way, that they would fit the cycle on the higher TF. Wouldn't they?

When we would have a deterministic universe, which would be based on Newtonian concepts, there clearly would be no probability, if we would know all conditions (Laplace's demon). But, as Hanover already explained a few postings ago, we are facing conditions/situations, which sensitivity to initial conditions exceeds our ability to measure them (or to know them).

So what now?
  • #12,926
  • Nov 20, 2014 5:24pm Nov 20, 2014 5:24pm
Quoting infinitus
{quote} So what now?
Simply let's trade as usual..
  • #12,927
  • Nov 20, 2014 5:30pm Nov 20, 2014 5:30pm
Quoting Neio
{quote} It was not a system as such but a way of looking at the market. Eurusdd did make a suggestion as to how it could be traded. The thing is you say that the Markov process fails. Maybe it does but the indicators that have been produced based on the theorem as presented show that you can configure h so that 97% of prices are h-recurrent and only 3% are transient. The problem I see that it is not a tradeable 97%...i.e most of the recurrence happens on the next bar or two and there is not enough gap between entry and the PTZ being cleared. I think...


Regarding the similarity thing:

As you may have noticed, I try to avoid to say system most of the time. Instead I say "thing".


Back to history:

EURUSDD gave you this similarity thing, or theorem, or whatever.

But since he never got tired to say, that he has far more knowledge, and only revealing part of it to us, AND claiming, that he would trade it, at least he must have a system (well, when we look at his busted trade explorer, well.....)


Fast forward:

the indicator produces a value, that shows 97% probability


OK, what does that mean?

Was it back-tested?

Which currency?

How big was the sample size?
How long was the tested period?
When was the testing period?

What was the average duration (or period) for price to reach similarity again?

What kind of draw-down do we have to expect?

Is the indicator kind of back-painting?

What has the indicator produced under pre-determinded conditions, like a stronger impulsive trend on higher TF, but the indicator placed on lower TF, or like an extended sidways correction on higher TF?


When the 97% are not tradeable, theindicator has failed. 97% should be tradeable, otherwise you should stop trading.
  • #12,928
  • Nov 20, 2014 5:31pm Nov 20, 2014 5:31pm
Quoting Luther
If you look at the 100% win ratio MT4 DEMO statement on the first post, the Closed Trade P/L is 10,627.29, but the Account Balance is 179.61, and Free Margin is 148.68. Why is that? I remember when the OP ran this DEMO Trade Explorer. He had the TE set up so Account Balance and Open Trades were locked, so the only information that was available were Closed Trades. When a Trade Explorer is set up this way, someone can hide losses on open positions while publicly showing a 100% win ratio on closed trades. An account Balance...

Exactly!!!!
  • #12,929
  • Nov 20, 2014 5:36pm Nov 20, 2014 5:36pm
  •  Neio
  • Joined Oct 2012 | Status: Can't think outside?Enlarge the box | 688 Posts
Quoting infinitus
{quote} Regarding the similarity thing: As you may have noticed, I try to avoid to say system most of the time. Instead I say "thing". Back to history: EURUSDD gave you this similarity thing, or theorem, or whatever. But since he never got tired to say, that he has far more knowledge, and only revealing part of it to us, AND claiming, that he would trade it, at least he must have a system (well, when we look at his busted trade explorer, well.....) Fast forward: the indicator produces a value, that shows 97% probability OK, what does that mean?...
The indicators where built to show the percentage of price (or bars) that is recurrent or transient for the configured h not to show what percentage is tradeable.
An expert is a man who has made all the mistakes which can be made
  • #12,930
  • Nov 20, 2014 5:41pm Nov 20, 2014 5:41pm
Quoting FranklinHood
@Kprsa How could this happen ?! There's a 10 pips TZ and your indicator says that 100% of TZ are < 2*k = 6 pips ! {image}
hi, well spotted, i need to check if it is a rounding issue or a bug. please note the value for the maximal zone size.
is this the latest version (4.2)?
k
  • #12,931
  • Nov 20, 2014 5:47pm Nov 20, 2014 5:47pm
  •  Neio
  • Joined Oct 2012 | Status: Can't think outside?Enlarge the box | 688 Posts
Quoting infinitus
{quote} Regarding the similarity thing: As you may have noticed, I try to avoid to say system most of the time. Instead I say "thing". Back to history: EURUSDD gave you this similarity thing, or theorem, or whatever. But since he never got tired to say, that he has far more knowledge, and only revealing part of it to us, AND claiming, that he would trade it, at least he must have a system (well, when we look at his busted trade explorer, well.....) Fast forward: the indicator produces a value, that shows 97% probability OK, what does that mean?...
No disrespect but I think you need to read the thread and explore for yourself if you are so inclined.

The indicators have a h that is configurable. So to achieve 97% you have to find the h for the given timeframe and currency pair.

Also it would be possible to find a h value so that almost surely every price is recurrent. Although probably useless as you would need to have a very high h value.
An expert is a man who has made all the mistakes which can be made
  • #12,932
  • Nov 20, 2014 5:49pm Nov 20, 2014 5:49pm
Quoting tashkent
{quote} i think there is a misunderstanding. i am not denying the possibility that one may make a long string of excellent trades using the similarity concept. i am questioning the REASON some people think behind that "high probable" events. that reason was explained by the TS (thread starter) as "arbitrary error" (sorry can't remember the exact wording). the question is WHY price must go back to that level? just market looked back and saw there was a "problem" in the past and went back to fix it? and we have seen some of those "problems" not fixed...


Great!!!

When we look at Volume at Price distributions, we can nicely see the areas, where market is trying to find balance. Often enough it then builds Gaussian or Gaussian like distributions.

When market decides suddenly to move, the balance is lost and price establish a new balance at a lower or higher level.

The room between the new balance and the old balance is most of the time only traded with small volume.

Often enough the low-traded area is retested again, or even the former balance zone.

But often enough, those zones are not re-challenged again for a very long time. Good luck in such a situation, when you bet on recurrence (or mean reverting)



Sigma 2 or sigma 3 events in Forex are not seldom. Even sigma 6 events we do see fairly often.

When you trade the similarity thing EURUSDD like, without SL, you will soon enough burn your account.


What made me wonder was the fact, that EURUSDD said, that he would use a 10 pip SL. But in all the trades I saw, he had not placed one.
  • #12,933
  • Nov 20, 2014 5:52pm Nov 20, 2014 5:52pm
Quoting Neio
{quote} No disrespect but I think you need to read the thread and explore for yourself if you are so inclined. The indicators have a h that is configurable. So to achieve 97% you have to find the h for the given timeframe and currency pair. Also it would be possible to find a h value so that almost surely every price is recurrent. Although probably useless as you would need to have a very high h value.

I know that the indicator has an "h"



so what was the "h" on the given TF's, which currency pair, which sample size and so on........
  • #12,934
  • Nov 20, 2014 5:58pm Nov 20, 2014 5:58pm
Quoting infinitus
{quote} When you do not define/describe the process you are looking for, how will you be able to calculate the right probabilities? Question: Will you bet a good part of the farm, when the probability is 97%? Will you bet the same, when the probability is "only" 80%?
Hi infinitus,

the probabilities are based on data alone, with no assumptions about the underlying process.
The code for the RecurrenceStatistic is free and open, and you are free to look inside.

The 97% probability is that a certain well defined event happens. The statistics do not calculate what happens in the 3% of cases or even the drawdown in the event of success. This indicator is not a complete strategy. The strategy and money management are up to the trader.

I trade pullbacks to strong trends, which is a common sense strategy. The indicator enables me to estimate the probability of trend proceeding (a higher timeframe fractal zone clearance) and to minimize the probability of hitting a stop loss (a lower timeframe fractal zone). It also allows me to estimate the reasonable trade lengths (time validity of trade).

k
  • #12,935
  • Nov 20, 2014 6:00pm Nov 20, 2014 6:00pm
Quoting tashkent
{quote} i have read Hanover's post and i do share the same thoughts with him. we all know that TS's ideas are really interesting and useful to some extend, no question. but i want to talk about his negative legacy. his view of the market helped some people here to shape not so healthy mindset like "beating the market" - to be one step ahead of the market. they all think that there are levels, specific numbers which price must reach. if price does not go the way they projected, they call it a "fake move". they developed a habit of challenging the...


fully agree

Maybe you know: I am an Elliott Wave trader (although combinig it with other methods). We Elliotticans think, that price is nearly exclusively driven by social mood. A change in social mood does not need an extrinsic factor. Moreover, it is also self-sustaining.

I find it pretty funny, when people say, price has to go to xxx in order to reach yyyy and so on
  • #12,936
  • Nov 20, 2014 6:02pm Nov 20, 2014 6:02pm
  •  Neio
  • Joined Oct 2012 | Status: Can't think outside?Enlarge the box | 688 Posts
Quoting infinitus
{quote} I know that the indicator has an "h" so what was the "h" on the given TF's, which currency pair, which sample size and so on........
Well your questions are not showing that you are understanding the purpose of the indicators.

Like I said before the indicators are showing the level of recurrency based on the user configured h. Most of the indicators allow you to specify the number of bars that you can include in the stats. It is all in the thread or obvious if you load the indicators and explore for yourself if you desire. It not I don't have the inclination to explain everything.
An expert is a man who has made all the mistakes which can be made
  • #12,937
  • Nov 20, 2014 6:03pm Nov 20, 2014 6:03pm
Quoting stt
{quote} just to correct this (as i many times corrected in EURUSDD's posts as well), in random market there is no predictability. you dont need to be math whiz to see this. His argument seems right when you have gamblers fallacy (i.e. if you think probability of Tail is higher after 10 Heads or as EURUSDD said, prob of one T in 7 toss higher after seeing 14 toss with all heads). In random market there can be no predictability - just think clearly and you will see why (hint: because random market means each price is independent of past moves).

do not know if one can say so:


m,aybe read about Wiener process, which is a Markovian process, or read about the success rate of Markov switching multifractal
  • #12,938
  • Edited 6:56pm Nov 20, 2014 6:13pm | Edited 6:56pm
Quoting Baillie
{quote} Here is the post you all are tripping over. Might I add; what a waste of time to keep on squabbling without reading the thread.

Nice, we have covered this point already.


I said a few postings ago, that I had notices about what EURUSDD said from earlier. The notices said, that EURUSDD thinks that similarity is Markovian. I made the notices before August, but can not find the postings anymore. Either deleted, or I only made an interpretation of what he said.


But all in all:

It is not really about what EURUSDD has said or not. Now he is eventually in the Asylum, has always only given part of information, has even used the indicator others have developed here, his logic was often enough flawed as well as his statistics.

It is about, if you like to know the reason, why most (maybe all) here will fail in the long run, when you believe in the 97% probability.

And the answer involves discussion about the right stochastic process.


Can you see that?
  • #12,939
  • Nov 20, 2014 6:14pm Nov 20, 2014 6:14pm
Quoting Kiads
{quote} Simply let's trade as usual..


That is a good answer
  • #12,940
  • Nov 20, 2014 6:17pm Nov 20, 2014 6:17pm
Quoting infinitus
{quote} how do you define sub-cycle? Is it simply cycle on lower TF?
Yes.

Quoting infinitus
{quote} Now we have to think for a while: When we believe, that the lower TF do follow the higher TF (a typical trader meme: lower TF must follow higher TF, because higher TF rule etc., you know this stuff all, I think) we have a problem: when we imply this, we would be living in a deterministic world. All events in the world would be needed to be designed in a way, that they would fit the cycle on the higher TF. Wouldn't they? When we would have a deterministic universe, which would be based on Newtonian concepts, there clearly would be no probability,...
I am personally not ruling out the possibility of somebody really clever knowing what the price signal will look like for eurusd in the next 24 hours. I do not possess this knowlege, but it is possible that somebody does.
Why do you perhaps use Eliott waves or statistical arbitrage? In order to be able to say with a degree of precision what is going to happen next, or what is not going to happen next. This is also just a method of technical analysis, and it is cleverly aiming at self-similarity.

I also happened to answer Hanover's post.

Now, I would appreciate if we would shorten this discussion.

k
7 traders viewing now