You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
It is a good idea to define a certain trading horizon as f(H,d) after which the DB will need revision. That's all for now.
There is one more thought, but you'll probably swear... If you go back to TF and look closely at the candlestick charts, the vast field of activity for the NS committee can be seen in a slightly different direction than we investigated before. I wrote you about it in your personal message, because the idea is probably a secret one - it is very simple and obvious.
It would be good to define some trading horizon as f(H,d) after which DB will need revision.
Let's be honest. The answer is obvious - the more often the better. I mean, ideally, as soon as it's full.
the database should be updated at a rate of computing resources. At the same time for each
for each stored pattern we need to recalculate probabilities taking into account the fresh history. It can be solved theoretically.
But if we look closely at the data gathering algorithm, here we have:
for (int i=0; i<CountPatterns; i++) {
1. we identify-fix the pattern.
2. Race the pattern through the history collecting statistics for it.
3. Write the result into the database.
}
Using the base:
1. We take a pattern from the market.
2. search for the most similar pattern(s) // synonym in Neutron's terms - the nearest node(s).
3. we play either (1) by the most similar-closest (2) by a Fuzzy-ified set of the most similar
(var: (2.1) by linear combination (2.2) by linear combination S(x), where S() is sigmoid, etc.)
Please, explain to me if I'm not an expert why the base is needed when I can gather statistics directly for the current market pattern. One of us is obviously stupid. I admit that I am. What's the problem?
:)
1. linear, polynomial, elastic (periodic) and other elementary-functional
There is no such thing as a linear polynomial, elastic, periodic, or other elementary-functional pattern in the market. None.
2. Neural networks are, however, useful, as they are able to capture the current poppy trade patterns.
3. Seized regularities are not long-lived and fade out gradually. 4.
4. At the input of NS it is better to have a variety of receptors.
5. Multicurrency basket play is twice as reliable as portfolio play.
6-7-8....poise.
3) Not. Discovered a dead end at the end of this (discussed) tunnel. Also a useful result, if you understand it.
4) Yes, thanks, removed. Didn't notice the message flying off twice.
1. On this point you said nothing.
2. We know.
3. Right.
4. Bullshit! A non-linear multilayer NS will build everything you never dreamed of from the input data by itself.
5. Friar! Can you point out the difference, in a nutshell, between playing the market with a multi-currency basket and with a portfolio of currencies?
Thanks for cleaning up the thread.
It would be nice to define a certain trading horizon as f(H,d), after which the DB will need to be revised.
At each interval the database needs to be updated. The effort is 0. The question is not worth the effort!
P.S. MetaDriver, says the same thing in his last post.
If we go back to the TF and look closely at the candlestick charts, the vast field of activity for the NS committee is seen in a slightly different direction than we explored before.
I'll have a look.
1. On this point, you said nothing.
2. We know.
3. Right.
4. Bullshit! A non-linear multilayer NS will build everything you could ever dream of from the input data itself.
5. Friar! Can you point out the difference, in a nutshell, between playing the market with a multi-currency basket and with a portfolio of currencies?
Thanks for cleaning up the thread.
1.2.3. OK.
4. Not nonsense. Just terminological inconsistencies, perhaps. The turkeys are also grids. The standard ones are basically
primitive single layers. However, are able to gather more information due to the "inclusive" coverage of bars over their
period. If you think about it, this is an advantageous feature. It can reduce the requirements to the NS superstructure.
5. The freak himself! ;) I can. But what's the point? Let's limit the definitions for the time being.
1) A portfolio (or a cluster) is a group of instruments of the following form
k[1]*С[0]/С[1] + k[2]*С[0]/С[2] + ... + k[n]*C[0]/C[n].....
where k are some coefficients (most often from the ceiling:),
C[0] - base currency (basis of the cluster)
C[1...n] - other currencies (C - Currency)
//definitions in terms of forex
2) Basket - multi-currency matrix N/N, or more exactly - common denominator (B - Basket) of fractions like
C[0]/B, C[1]/B,..., C[n]/B in such matrix.
As for prediction strategies-models, they may be different, but it doesn't change the diagnosis.
As a rule, portfolios are played (as a rule along the supposed trend) to decrease the risk of trend reversal. However
risks are half as much if we play in two counter-trends. I think it's obvious. The probability of two
of two counter-trends is obviously exactly half the probability of one. And regardless of the absolute value
of the probabilities. No? ;)
I need to think about it.
The probability of two reversals is clearly exactly half that of one. And irrespective of the absolute value
of the probabilities. No? ;)
Nah! If there were Martian quid trading in the market, imho, you'd be right. However, where to get two different trends in the same market is beyond me.
Nah! If the market was trading masian quid, then imho, you'd be right. However, where to get two different trends in the same market is beyond me.
Catch up already.
One trend on C[i]/B, the second on C[j]/B. Let's play, of course, on C[i]/C[j]. Huh?
I.e. an approximate algorithm:
1. Count the basket.
2. Sort currencies by trends.
3. Play by the furthest in the sorted list.
3-a. If you like, you can also take the second and the penultimate one.
4. Pray that ALL tendencies do not unfold at once. Option: We drink beer in peace because 4 is supposedly unlikely.
Okay? :)
Catch up already.
One trend at C[i]/B, the other at C[j]/B. We play C[i]/C[j], of course. >> huh?
It's a cross. Getting into trouble on a game like this is like getting your feet wet, and no amount of stopping will save you.
It's a cross. Getting into trouble on a game like this is a no-brainer and no amount of stopping is going to save you.
I didn't make any guarantees. I was just looking at the odds. No more than that.
But keep in mind, probability calculations are in vogue these days. ;) :) :)
I've seen this fashion... you know where. I need profits and only profits, and I don't care if I take them in a fashionable way.
Of all what I have tried in real trading, so far only one tactic has proven effective (not a strategy, mind you):
1. Limit the losses.
2. Limit the losses.
3. Limit the losses.
It's not trendy, but you don't have to pray while drinking beer - which, by the way, is also not useful.