Is any prediction doomed? - page 14

 
moskitman:

The franc is highly oversold now. Moreover, the franc is oversold so much, that I wouldn't make conclusions about buying it on Monday, because its behavior doesn't fit any economic considerations. Personally, I would prefer to wait out of the market, and then on Tuesday-Wednesday, we'll "look and see". But this is MY analysis, MY forecasts and MY vision of the market, the results of this analysis with all their profits and losses will also be mine.
This is about something else. Multicurrency analysis and, as a consequence, forecasting the behaviour of several instruments actually has a basis in purely economic considerations, aligning at certain points in time fundamental movements, which is exactly what we are seeing in the behaviour of the franc right now.
Again, don't be so categorical...

Understand, you are essentially proposing to analyse each instrument individually for overbought/oversold, so what do multicurrency "alignments" have to do with it? They can go unaligned for years at all.

Yes, a correction will start sooner or later, but it will also happen for each instrument separately, although, at the same time, for all of them at the same time.

The question is, what is the difference, whether we will detect a reversal on one pair or on several at once? What is the advantage? Apart from unnecessary additional loading of the depo, I do not see any rational reason.

 
The trade assumes that the results of the system are quasi-stationary, i.e. stationary temporarily and with some error. Probability estimates then have some meaning. Such systems are called robust systems
 
-Aleksey-:
Wrong. The degree of non-stationarity is measurable, and this allows one to correct the probabilistic estimates by making them meaningful. The proof of their meaningfulness in the non-stationary case, by the way, is constructed by means of the main laws of classical TV.
Proof of correcting what? The non-stationarity of the market? Well, well, well, I am interested to see how such meaningfulness is proved for the process, which mathematicians can't even describe properly.
 
HideYourRichess:
Evidence of adjustment to what? The non-stationarity of the market? Well, well, well, I'd be interested to see how such a proof is made for a process that mathematicians can't even describe properly.
Avals, right answer. Mathematicians described it all long ago. Certainly not in textbooks.
 
-Aleksey-:
Avals, correctly answered. Mathematicians described it all long ago. Certainly not in textbooks.
I, a little mathematician in a former life, don't need to be told that, I have an approximate idea of what we are talking about. Anyway, it's all nonsense in relation to the market. The market will surely punish you for it. ;)
 
HideYourRichess:
Proof of adjustment of what? Non-stationarity of the market? Well, well, well, I'd be interested to see how such a proof of content can be found for a process that mathematicians can't even describe properly.


All right, enough with the quibbling. Let's get back to the main issue - the gap between finding patterns and predicting.

You have found a pattern, how can you use it to trade without predicting? How? How do you "just use it"?

"Generally speaking, predicting in financial markets is a useless business. Only searching for patterns....." - I do not understand this phrase at all. Isn't the forecasting based on finding and exploiting the patterns found?

 
OnGoing:

Understand, you are essentially proposing to analyse each instrument individually for overbought/oversold, so what do multicurrency "alignments" have to do with it? They can go unaligned for years at all.

Yes, a correction will start sooner or later, but it will also happen for each instrument separately, although, at the same time, on all of them at the same time.

The question is, what is the difference, whether we will detect a reversal on one pair or on several at once? What is the advantage? Apart from unnecessary additional loading of the depo, I do not see any other rational reason.

Well, if you don't see it, you just don't want to see it. I've traded extra depot load for reliable trading signals and I'm fine with that. I have previously abandoned the notion of "detecting a reversal" as a one-currency atavism. If I see a "tension" created by the foundation in the market, I enter the market with a big amount of orders to its liquidation. By this time I already know who "wants" to go where, which is usually enough for a positive order basket balance.
 
FAGOTT:


OK, enough of the quibbling. Back to the main question - the gap between finding a pattern and predicting.

You have found a pattern, how can you use it to trade without predicting? How? How do you "just use it"?

And how do you go about crossing the line? Everything has already been said, you only need to think, and agree or disagree - it's up to you, no one is going to impose an opinion (I'm talking about me), all the money and all the risks are yours alone.
 
FAGOTT:


OK, enough of the quibbling. Let's get back to the main issue - the gap between finding patterns and forecasting.

You have found a pattern, how can you use it to trade without predicting? How? How do you "just use it"?

"Generally speaking, predicting in financial markets is a useless business. Only searching for patterns....." - I do not understand this phrase at all. Isn't the forecasting based on finding and exploiting the patterns found?


The question is probably about distinguishing patterns from pseudo-laws. Like the classic example of teachers of statistics - the correlation between the number of babies born and the stork population :)
 
Exactly.