Application of mathematical analysis and higher mathematics - page 12

 
Tovaroved писал (а):
Mathemat wrote (a):

Why no one... The complete randomness of price behaviour (Brownian motion) is only one hypothesis about the market, and not a very good one at that. There are significant levels, Elliott waves etc.

Elliott waves are a shitty formalisation that is further confused by gore analysts...

in fact, these waves have much deeper properties (and much more specific and definite than one might assume),
or rather, all these levels and waves are just the visible part of the iceberg, just the consequence of an inherently elementary process,
to understand which you don't even need a special mathematical education.
(I do not know much myself, but I know that the process can be described algebraically and geometrically with simple equations.
But by such peculiar banter I try to give people some food for thought :) )
Not much constructive, MTS does not involve gore-analysts, any suggestions?
 
FION писал (а):

Not much constructive, MTS does not involve gore-analysts, what are the suggestions?
I don't get it. What do you mean?
I suggest we use the special theory of relativity, Minkowski's geometry and Descartes' mechanics to build a flying saucer. :)
 
Tovaroved писал (а):
FION wrote (a):

Not much constructive, MTS doesn't involve gore-analysts, any suggestions?
I do not understand. what do you mean?
I suggest using the special theory of relativity, Minkowski's geometry and Descartes' mechanics to build a flying saucer. :)
Or... a lot of noise out of nothing. Algebraic and geometric simple equations, please!
 
Tovaroved писал (а):
What is noise for a long term pipser, is mad dough for a pipser!

That's a good one. I am of the same opinion. Many analysts try to get rid of high-frequency price changes, considering them as noise. That is why muves were created, which are essentially digital filters. Their task is to simulate price behaviour on longer timeframes. But everyone understands that smoothing of the curve reduces its length, reducing the number of trades. For traders, high-frequency price fluctuations are very important to maximise profits. For me, the more volatility, the better. What long term traders see on daily timeframes, I see on minute or hourly timeframes, which means that theoretically the same money is made much faster (this is the goal of all traders).

Of course the most important question is how to predict the high frequency price behaviour. Slow price behaviour (e.g. over the course of a year) can somehow be correlated with economic predictors. But how do you predict the behaviour of prices during the day? We should first evaluate how random the price behaviour is on small timeframes. This will determine the choice of model. Suppose the behavior of prices is completely random. That is, at any moment the price can go either up or down with the probability of 50%. The step size is equal to one tick. This is a classic random walk. Can one make money on such a market? Of course it is. But a mathematician will need a lantern here. Trading on such a market is like driving on a winding mountain road at night: you cannot see the turns in advance and predict them. Nevertheless, when the road turns, after a slight oversteer (stop-loss), it is possible to adjust the direction of the car so that it returns to the road. The whole idea of such trading comes down to the assumption that, from time to time, price will move in the same direction for longer than one timeframe. Then our losses on stoplosses will be compensated by gains on the "straight road". Most indices are precisely invented with the assumption that price behaviour is completely random. To use my analogy with the road, the indices act as a central marking on the road that allows you, first, to establish the fact that the road turns and second, to estimate how far the road turns from the straight.

Now let's assume that the market isn't entirely random. I am one of those who think that there is a certain regularity in the behaviour of prices. The question is how to use this pattern in predicting prices. Here mathematical models are very applicable. Neural networks, for example, do not seek to understand that pattern in prices. They simply teach the Expert Advisor to make profitable transactions under certain circumstances that have led to profitable transactions in the past. That is, it is cloudy in the morning, take an umbrella because experience shows that the probability of rain is high. It does not matter why it will rain. Other mathematical methods try to fit some kind of model to the behaviour of prices. The choice of models is wide: each trader tries to draw an analogy between price behaviour and something more familiar from work, for example, the behaviour of an unstable multivibrator or gas diffusion. The market is a black box with many unknown inputs and an unknown internal structure. The only thing we know about the market is the behaviour of prices. It is our output signal. The internal structure of the market, as some have pointed out here, is non-linear. That is, the input signals are distorted. For example, news of the same nature causes different-size-and-character price movements. I work in the field of electronics. Therefore, the analogy between the market and a non-linear circuit affected by pseudo-random signals (e.g. communication signals) is closest to me. There are many methods to analyse such circuits and their output signals. For example, Fourier transform, Volterra series, sampling theorem, wavelets, etc. There are a lot of methods, and we have little time to thoroughly research them for forex trading. But there is nothing to do. The road is travelled by those who walk.
 

Sergei, there is no noise. and the equation will do nothing for anyone.
And I don't know enough about it to try to explain it, and I don't see the point in talking nonsense.
And what I can articulate, I've already said. I'm just stating that it's quite possible to figure it out, and it won't take a lifetime if you work regularly...

What degree of generalization is the equation? The most general? Then y=n/x
That's one of the options. What are the others? I don't know.

And if you want some food for thought, here's one...
http://www.trinitas.ru/rus/doc/0232/004a/02321053.pdf


 
gpwr писал (а):
Tovaroved wrote (a):
What is noise for the long term, for the pipser it's mad dough!
Of course, the most important question is how to predict high frequency price behaviour. Slow price behaviour (e.g. over the course of a year) can somehow be correlated with economic predictors. But how do you predict the behaviour of prices during the day? We should first evaluate how random the price behaviour is on small timeframes. This will determine the choice of model. Suppose the behavior of prices is completely random. That is, at any moment the price can go either up or down with the probability of 50%. The step size is equal to one tick. This is a classical random walk.

Now let's assume that the market is not completely random. I belong to those who think that there is a certain regularity in the price behavior. The question is how to use these regularities in price prediction.
There are a lot of methods, there's not enough time to study them thoroughly for forex trading. But there is nothing to do. If one walks the other the way is the right one.

The paradox is that this same random walk 50/50 can be mathematically described, and then it becomes completely predictable. That is, you can predict the future behavior of a chart created in Excel using the RAND() function... You don't believe it? Neither do I. But facts are stubborn.


Yeah, well... that's an awful lot of time.... I'm off to work... happy holidays until January... hopefully I'll be back here again...
 
Tovaroved wrote:

The paradox is that this random walk is 50/50 mathematically describable, after which it becomes completely predictable. i.e. you can predict the further behaviour of a graph created in Excel using the RAND() function... You don't believe it? Neither do I. But facts are stubborn.
Random walk is unpredictable in the classical sense. You can calculate its parameters, but you can only predict where the random walk will be in 10 periods with a very low accuracy. If you have it predictable, then you are exploiting the non-randomness of the RAND() function.
 
Mathemat писал (а):
Tovaroved wrote:

the paradox is that this same random walk 50/50 can be mathematically described, after which it becomes completely predictable. i.e. you can predict the further behaviour of a graph created in Excel using the RAND() function... You don't believe it? Neither do I. But facts are stubborn.
Random walk is unpredictable in the classical sense. You can calculate its parameters, but you can only predict where the random walk will be in 10 periods with a very low accuracy. If you have it predictable, then you are exploiting the non-randomness of the RAND() function.

I join the mathematician. RAND() actually generates a pseudo-random sequence of numbers. Tavoroved, if you can actually predict a random walk then you should apply for a Nobel Prize. If I flip a coin, then you are essentially claiming that you can predict the result of such an experiment with more than 50% probability by looking at previous results. I claim that the result of such an experiment would be heads or tails with a 50% probability with every flip. If you are right, why don't you write a lottery number guessing advisor?
 
Tovaroved писал (а):

And if you want some food for thought, here's one...
http://www.trinitas.ru/rus/doc/0232/004a/02321053.pdf


Do you really use this somehow in forex or is it just for fun ?
 
Yurixx:
Tovaroved wrote (a):

And if you want some food for thought, here's one...
http://www.trinitas.ru/rus/doc/0232/004a/02321053.pdf


Do you really use this somehow in forex or is it just for fun ?
You might as well use torsion field theory.