Home › Forums › General Discussions › The similarity system – discussion
Tagged: calculating probabilites, market DNA, similarity, transient
This topic contains 246 replies, has 23 voices, and was last updated by Anti 5 months, 2 weeks ago.

AuthorPosts

But what do you mean be applying standard deviation instead of correlation? SD of sochastic?
Bollinger Bands are based on standard deviation. So when applying BB, you always consider standard deviation. That’s what I meant.
I interpreted your proposal in a way that you would replace the correlation approach by BB. Thus correlation would be replaced by standard deviation.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top and bottomfishing to people on an ego trip. (Dr. Alexander Elder)
Ah. No. I’d like to put BB with identical settings on prize as well as stochastic and use those BB levels (SMA, +2 sd, 2 sd) to identify deviations between prize and stochastic. This is similar to what Ralome’s indicator does (but maybe with some refinements) …
Thanks, got it!
All in all the approach reminds me of my years of (very) early adulthood. Then we were having fun giving our friends’ automobiles names like ‘pothole detector’. A pothole detector worked best with a chassis suspension trimmed tougher than average, some call this a ‘sports chassis’. So an ambigiuos feature that was a perceived as a lack of comfort by most and a must have by ‘sports’ oriented drivers could be utilized to count the potholes in the road.
Here the mathematical inefficiency of the Stoch algorithm (its hard limits in the extreme zones) is about to be exploited to detect potholes in the road of prices.
I’m not so sure whether this will work, but I’m really interested to study some significant statistics about the approach.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top and bottomfishing to people on an ego trip. (Dr. Alexander Elder)
Will give my very best. Unfortunately, atm I have too much work to do (PhD, preparation of biostatistics course, student assistant job, summer schoool, …)
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top and bottomfishing to people on an ego trip. (Dr. Alexander Elder)
On the other hand: after the 3rd event in your chart pic, you can see an obvious divergence between price and Stoch(300): Stoch slowly decreasing, while price increasing further. Good old Stoch ambiguity working against us!
Yes, that’s a perfect example where I can present that my modified stochastic doesn’t show a divergence:
The reason is simple: the usual stochastic shows a divergence due to a range compression as candles with lower lows subsequently leave the lookback window %k. My modified version doesn’t care. Thus, it show convergence with prize …
Nice one! #AntiModStoch looks like an interesting approach.
But now the combination of standard deviation (which period did you use for BB?) and long period Stoch range compression is playing tricks on us (see that very sudden drop to the base line below the ‘o’ of ‘modified’).
‘Zu Risiken und Nebenwirkungen lesen Sie die Packungsbeilage …’
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top and bottomfishing to people on an ego trip. (Dr. Alexander Elder)
`… oder fragen Sie Ihren Arzt oder Apotheker.´
There’s still no BB. It’s just the version I introduced <a href=”http://penguintraders.com/forums/topic/thesimilaritysystemdiscussion/page/2/#post12758″>here</a>. But yes, those strong crosses above/below the 50 level are negative side effects which one has to ignore or use to suggest trend changes for bigger %K_mod.
There’s still no BB.
Now I’m confused. I thought your mod is a combination of BB and Stoch.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top and bottomfishing to people on an ego trip. (Dr. Alexander Elder)
No, the mod version is based on modified Donchian Channel which only plots new lower levels if usual Donchian Channel plots new low levels, and vice versa for upper level. The BBStoch or BBStoch_mod was an idea to spot potential reversals regarding 3plan.
Sorry if that was too confusing.
 This reply was modified 2 years, 8 months ago by Anti.
Sigh … My advanced age, you know …
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top and bottomfishing to people on an ego trip. (Dr. Alexander Elder)
I am agree with @simplex, there is no such “optimal setting” , i think we must spend our time to try, try, try and try, looking for the best setting for each pair, that work on specified TF, under some conditions like trading session time (asian, europe, us), our trading style, avoid news, kapiere, etc … so, there is no setting that can match all conditions …
Thanks for the reminding of old stuff @anti, i am sure that it will Certainty support our understanding on how the price move based on the simple indicator we have, may be we just need to apply “trying and error” method to get better and better result. As kprsa said, all are related (MA, Stoch, ZZ, Cycle, Channel, etc). Regarding the “flat” line of MA, lets say we have “nearest best setting value” of MA we can get, so we can see the line “almost flat” on TF H1. In this case can we assume that it represent the “resonance line” of the swing of a pendulum ? Furthermore, may be we can use Heiken Ashi or other stuff (like BB and its deviation) instead of MA too ? In case of using Excel to draw the line, after we export the OHLC of each candle data to CSV file from MT4 for example, import it to Excel, do we need to create a macro rule in Excel to draw the line, or is there any idea how to do it ?
@smallcat: You’re resonance analogy may be good in that case.
I’m not really sure if there are other simple ways to identify current representative cycle period (CRCP). The problem is simply that our CRCP may change with every new candle. Thus, other approaches like finding best %K for which the stochastic bounces less back to the 0 or 100 level are worthless. No, I haven’t any idea how to do it, either in MT4 nor in Excel.
Ok folks, until I’ve more time to do some tests on stoch, BB, ZZ, and prize, I’d like to discuss further topics. Maybe this one is the most basic one:
As far as I understand it, we deal with two stochastic processes (X,µ,d) on X1 (prize) and X2 (indicator values). But what exactly does that probability statement (1.1) mean and what conclusions can be drawn from it?! Any ideas/suggestions?
And how can that lambda be interpret? Eurusdd mentioned that it serves as waiting time … But how exactly?
 This reply was modified 2 years, 8 months ago by Anti.
Would you support to interpret those events of correlation dropping below zero as a nonrepainting alternative to ZigZag, indicating major points of reversal?
No, it’s not. You can clearly see that a the major leg is associated with a correlation above zero but repainted legs partially show higher degree of anticorrelation.
Think about what Eurusdd said: Whenever the stoch is above/below the critical level and no ZZ occurs afterwards, we have dissimilarity which must be corrected. If we have the right setting, in 98% of the time there should always appear a ZZ leg after stochastic was at critical level …
 This reply was modified 2 years, 8 months ago by Anti.
Ok folks, until I’ve more time to do some tests on stoch, BB, ZZ, and prize, I’d like to discuss further topics. Maybe this one is the most basic one:
As far as I understand it, we deal with two stochastic processes (X,µ,d) on X1 (prize) and X2 (indicator values). But what exactly does that probability statement (1.1) mean and what conclusions can be drawn from it?! Any ideas/suggestions? And how can that lambda be interpret? Eurusdd mentioned that it serves as waiting time … But how exactly?
Thanks for the great post Anti Waiting time … hm … interesting, but do not know what Eurusdd meant about it in term of Similarity …
Maybe I waste my time over and over again. However, I’m still studying some of the concepts of Eurusdd. Although they seem to be all very interesting and logically correct, I believe that some parts are missing to be successful. For instance, the probability statement discussed here only seems to be valid if we are looking at chunks of candles and if these blocks of 5 consecutive candles do not overlay.
In order to find some combatants and to start comprehensive discussions on the various topics, I decided to share a compendium of all (important) posts of Eurusdd.
Eurusdd mentioned many times that prize will revisit prize areas where dissimilarity started. Those dissimilarities are market inefficiencies. But what are those inefficiencies? What do they mean? And why do market makers have to correct it (although they sometimes seem to forget about it …)?
https://youtu.be/knQ_ZqpmzC8?t=947
chris lori has the same view on inefficiencies as i do. he also tries to explain why gaps happen.
i know this isnt ss or indicator based sim divergence but i believe it is still relevant to the original post
“I believe the very best money is made at the market turns. Everyone says you get killed trying to pick tops and bottoms and you make all
your money by playing the trend in the middle. Well for twelve years I have been missing the meat in the middle but I have made a lot of
money at tops and bottoms.”
– Paul Tudor JonesHi @lowphat,
will have a look on it. Thank you very much!
ATM I read very much on TZs and subsequences. Now I stumbled upon an interesting sidebehavior of PTZs which may give a simple hint on when a retracement or a reversal may happen. Assume that you use PTZs wih h bars to both sides. Whenever we see a righthand potential TZ but the next bar, which has a higher high or a lower low, negates that PTZ, then we can assume that either a retracement or a reversal may happen. Mabye most of you have still encountered that behavior, but for me it was a new noob observation. Surely that isn’t true whenever the PTZ forms around a long wich of a hammerlike candle …
Ok, back to Eurusdds formalization of the similarity principle:
As far as I understand it, we need a function
g(a)
which maps every unique prize levela
(e.g. close prize) to another valueb
. To speak of λisomorphic setsA
(a ∈ A
) andB
(b ∈ B
), it must be possible to conclude from a particulara
to a particularb
and vice versa (bidirectional mapping). Eurusdd mentioned that all methods discussed in the similarity thread, don’t identify real dissimilarity zones. To show my explanation on an example, I’ll refer to the StochasticBollingerBand method: Here, the prize itself and the stochastic values are not λisomorphic as it is possible to conclude a corresponding stochastic value from each prize level , but it is not possible the other way around if stoch=0 or stoch=100. (That is also the reason why I developed the modified stochastic discussed here – but I’m not sure if it really solves the complete probleme).Thus, the question first should be: Which function could generate a λisomorphic set to the prize set and still produce some dissimilarities from time to time?!
Comments and discussions are requested (but no longer expected …).
Thanks for the great post Anti Waiting time … hm … interesting, but do not know what Eurusdd meant about it in term of Similarity …
Maybe similarity and TZs are two practical applications of the same theoretical concept. Thus, λ could be either the waiting time from dissimilarity to similarity or from dissimilarity to another dissimilarity (which is more plausible to me).
 This reply was modified 2 years, 6 months ago by Anti.
Pretty much. Gap, Wormhole, PTZ, inefficiency . Something that the market will soon correct. When to enter, what to risk, when to exit?
Past performance does not garuntee future success. You still have to manage your trade. No way to shake it.
EURUSDD explained 10 PTZs per day on average will be corrected given your settings for h and the probability of it clearing. You want a 97% probability or better he explained.
He used the phrase “Almost surely” quite a bit. Means no garuntee.
In the end it’s not safe as is.

AuthorPosts
You must be logged in to reply to this topic.