The similarity system | Page 566 | 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

  • #11,301
  • Oct 30, 2014 8:52am Oct 30, 2014 8:52am
Quoting NorthTrader
{quote} No, I meant where h is set such that 97% of bars on the chart are recurrent (either left-only, right-only or fully recurrent), even if the chart contains 23 million bars of data. In that siutuation, the probability of a potential TZ bar becoming resolved (i.e. right-only recurrent) can't be 97%, can it? It must be less, since "right-only-recurrent" is a sub-group of the 97%. In fact, I'm speculating that it will be closer to the 79% figure that I obtained from my small sample size.
I'm beginning to think you are right. Thankyou for your insistance!
I want more brain power!
  • #11,302
  • Oct 30, 2014 8:55am Oct 30, 2014 8:55am
The idea of similarity that I presented here is simple for a lot of people to understand. But that is not real similarity. The real idea is rigorous and beyond the scope of most traders here.
However, if you really get it you will realize that just the way two distinct algebraic systems can be isomorphic, two random systems can also be isomorphic, almost surely.

How you construct such systems is the secret of isomorphism applied to trading.
There will be bars where both systems do not agree but the measure of the set of those bars will be ZERO!!!!

fOR EXAMPLE, the following sequence of positive integers is randomly isomorphic to a psedu-random sequence which is well-known. The set of dis-similarity is empty, so of measure zero. Yet if you meet this sequence for the first time and even after studying it for years, you may never be able to predict the next term with certainty.

6-7-9-11-15-17-21-23-33-35-41-45-47-51-57-61-65-71-77-??????

For random systems, the set of dis-similarity may not be empty or of measure zero: you have to quantify that set to have a secured similarity system. Too technical to explain here so people should just stick to the basic ideas.

For those interested we can communicate privately.
  • #11,303
  • Oct 30, 2014 9:09am Oct 30, 2014 9:09am
i return, we will go back to making calls and testing our ideas. bye.
  • #11,304
  • Oct 30, 2014 9:14am Oct 30, 2014 9:14am
Quoting Eurusdd
6-7-9-11-15-17-21-23-33-35-41-45-47-51-57-61-65-71-77-??????
83

Quoting Eurusdd
For those interested we can communicate privately.
interested

k
  • #11,305
  • Oct 30, 2014 9:17am Oct 30, 2014 9:17am
Quoting FX-Jay
{quote} The 79% figure you have is quite accurate. It's the 79 that is trad-able, not the 97. This is what most folks around here are not getting. 79% probability is pretty good, but not good enough to be real profitable. That's why the trick here is how to make this 79% increase to 90+ There's more than one way to do that, including increasing the H value, or playing with the K value. There are some other ways to do so that are not scientifically proven/valid, but are working more than well for me. ​I used to worry to write these out loud,...
Thank you Jay. What's more, if that 79% probability is accurate, then it follows that the probability of price returning to any bar on the chart (PTZ or not) is always close to 80%. I can prove that tomorrow, but it's late now in Japan and I need to get to bed. This assumes of course that h is set such that 3% of all the bars contain full TZs.

Yes, you can improve the probability by increasing h. For instance, doubling h-right increases the probability to about 85%, and quadrupling h-right increases it to about 90%. I can go into details tomorrow.

I was wondering how you managed to give probabilities of more than 90% in your live trade calls, so your hints are very interesting! Thanks again.
You reap what you sow.
  • #11,306
  • Oct 30, 2014 9:30am Oct 30, 2014 9:30am
Quoting Eurusdd
The idea of similarity that I presented here is simple for a lot of people to understand. But that is not real similarity. The real idea is rigorous and beyond the scope of most traders here. However, if you really get it you will realize that just the way two distinct algebraic systems can be isomorphic, two random systems can also be isomorphic, almost surely. How you construct such systems is the secret of isomorphism applied to trading. There will be bars where both systems do not agree but the measure of the set of those bars will be ZERO!!!!...
I'm very interested in figuring this out, happy to communicate privately (your PM is off).
  • #11,307
  • Oct 30, 2014 9:42am Oct 30, 2014 9:42am
Quoting kprsa
{quote} 83 {quote} interested k
You are right!!! The sequence I provided is exactly randomly isomorphic to the sequence of prime numbers. I have not defined what "randomly-isomorphic " means precisely but i believe most of you can see through this.
Now, suppose you do not know the sequence of prime numbers, you will never know that there is a simple function that can show the relation.

Look at the following example: The sequence of prime numbers is pseudo-random, very uncertain in its distribution but do you know that despite this uncertainty, given any integer K, there is a prime in the interval [k, 2k]?

So, yes we may not know exactly where the primes are, but we can know how long to wait to HIT ONE where the sequence of integers develop.

The same applies to trading. I can defined functions that can link me to eurusd IN A WAY THAT ANYTIME my function takes a value, I know how long to wait for eurusd to respond in a certain way. More privately later.
How I construct these functions is secret but I am finalizing my blueprint for after december when I am back to trading!!!
  • #11,308
  • Oct 30, 2014 9:58am Oct 30, 2014 9:58am
Quoting Eurusdd
i return, we will go back to making calls and testing our ideas. bye.
yay!! looking forward to it. you have unique way to analyzing the markets. interesting in discussing privately too - think i can add value but of course you decide :-)
Think : Trade : Live Life
  • #11,309
  • Oct 30, 2014 10:01am Oct 30, 2014 10:01am
Quoting Eurusdd
{quote}You are right!!! The sequence I provided is exactly randomly isomorphic to the sequence of prime numbers. I have not defined what "randomly-isomorphic " means precisely but i believe most of you can see through this. Now, suppose you do not know the sequence of prime numbers, you will never know that there is a simple function that can show the relation. Look at the following example: The sequence of prime numbers is pseudo-random, very uncertain in its distribution but do you know that despite this uncertainty, given any integer K, there...
I am blown out of the water again like a duck during duck-hunting-season.
A pip is only worth it if you know how much you risked to earn it
  • #11,310
  • Oct 30, 2014 10:12am Oct 30, 2014 10:12am
Quoting kprsa
{quote} 83 {quote} interested k
Would it be too much to ask how u arrive at 83?
  • #11,311
  • Oct 30, 2014 10:26am Oct 30, 2014 10:26am
Quoting summicron
{quote} Would it be too much to ask how u arrive at 83?
Eurusdd's example was (approximately) the prime number sequence shifted by 4.

Attached Image


k
  • #11,312
  • Oct 30, 2014 10:34am Oct 30, 2014 10:34am
Quoting kprsa
{quote} Eurusdd's example was (approximately) the prime number sequence shifted by 4. {image} k
K you are Da Bomb!
A pip is only worth it if you know how much you risked to earn it
  • #11,313
  • Oct 30, 2014 10:44am Oct 30, 2014 10:44am
Quoting kprsa
{quote} Eurusdd's example was (approximately) the prime number sequence shifted by 4. {image} k
you've got a bright mind!
however when trying to figure it out,
that gap between 23 and 33 fooled me a bit...
Mind is the King
  • #11,314
  • Oct 30, 2014 10:45am Oct 30, 2014 10:45am
Quoting kprsa
{quote} Eurusdd's example was (approximately) the prime number sequence shifted by 4. {image} k
He missed 27 in his sequences
Quote
6-7-9-11-15-17-21-23-33-35-41-45-47-51-57-61-65-71-77-??????
  • #11,315
  • Oct 30, 2014 10:55am Oct 30, 2014 10:55am
Quoting FX-Jay
{quote} one way to do that, including increasing the H value, or playing with the K value
That is the right way idea.

One thing to think of - what if we let us have the real probabililty to be equal 1 is not for ONE timeframe, but for all TFs included.
Hunting High and Low
  • #11,316
  • Oct 30, 2014 10:56am Oct 30, 2014 10:56am
Quoting xixi
{quote} He missed 27 in his sequences {quote}
I am quite happy with 90%+ precision.

Cheers,
k
  • #11,317
  • Oct 30, 2014 11:03am Oct 30, 2014 11:03am
Quoting kprsa
{quote} I am quite happy with 90%+ precision. Cheers, k
  • #11,318
  • Oct 30, 2014 11:06am Oct 30, 2014 11:06am
Quoting kprsa
{quote} I am quite happy with 90%+ precision. Cheers, k
Yes. But that is part of the set of measure zero. For a certain number of times your system and mine may not agree but it is clear you know what sequence i was looking at. So, good work done. That is the spirit of the issue.
  • #11,319
  • Oct 30, 2014 11:36am Oct 30, 2014 11:36am
Quoting NorthTrader
{quote} Thank you Jay. What's more, if that 79% probability is accurate, then it follows that the probability of price returning to any bar on the chart (PTZ or not) is always close to 80%. I can prove that tomorrow, but it's late now in Japan and I need to get to bed. This assumes of course that h is set such that 3% of all the bars contain full TZs. Yes, you can improve the probability by increasing h. For instance, doubling h-right increases the probability to about 85%, and quadrupling h-right increases it to about 90%. I can go into details...
NorthTrader is probably well aware of this from reading my blog but this is basically my issue as well. Perhaps it depends on how one views the propositions when they are first introduced to them, but trading that 80% is what I focused on quite a bit. One of the "problems", as illustrated by the time/probability scale by NorthTrader is that your probability to hit decreases rapidly as time increases, and you get less "bang for your risk" I suppose. So to me it's really the k value that's the important piece here, which maybe if I'm lucky jay will let the cat out a little more =p


Quoting Eurusdd
The idea of similarity that I presented here is simple for a lot of people to understand. But that is not real similarity. The real idea is rigorous and beyond the scope of most traders here. However, if you really get it you will realize that just the way two distinct algebraic systems can be isomorphic, two random systems can also be isomorphic, almost surely. How you construct such systems is the secret of isomorphism applied to trading. There will be bars where both systems do not agree but the measure of the set of those bars will be ZERO!!!!...
Who knew math could be so interesting? I would definitely love to know more
"Mrrraggglhlhghghlgh"
  • #11,320
  • Oct 30, 2014 12:07pm Oct 30, 2014 12:07pm
For those using TradingView (web charting platform) might be interested in this little TZ indicator I coded in a few minutes.
It's really limited but helps to identify TZ and PTZ a lot faster.
Good enough when MT4 is not available for whatever reason...
Script on TW (https://www.tradingview.com/v/X9fqF4cX/)
Attached Image (click to enlarge)
Click to Enlarge Name: tz_tw.png Size: 65 KB
the 3rd wave is the crowd's wave
1 trader viewing now