A good mathematician should be good in forex. - page 4

New comment
 
erdogenes:

  • Mandelbrot (if I do remember truely the date is 1960) claims the price returns dont conform the central limit theorem, he told -by his research- this is a known issue since early 1900s but financial mathematicians ignore the findings because there is no other way forward. and they still use the normality axiom. and develop their models and methods accordin to normality axiom.  the real distributions of financial data (which is stationarized) is not distrubuted normally nor aproximate to normal. they are called heavy-tailed distrubutions.

CLT is mathematically formed. And it is proved in more then one way. So it is correct.
Mathematics is very exact. To prove that a theorem is false, you just need to specify one case for which it isn't true. I didn't see such prove.

So this is a matter of false interpretation. What does CLT say?
It says "bla bla bla bla  when n -> infinity bla bla bla".

So the CLT doesn't claim that you will have normal distribution on any sub-domain in time (ok, domain defined by time period), but that it will have normal distribution when number of samples goes towards infinity. 

I don't know which distribution type takes lets say €$ in a year, as i didn't check it (but i will). However, i have checked it for stationery. And it is non-stationary on a year to year basis.

And i have found claims that a random walk tends toward normal distribution, what can be easily checked out. So, logically this should be also true for PA.

However, i don't see a way to build a profitable model on the idea that distribution tends to be normal. 

 

  • why  is a distribution is so important: because there is no standard way to proof financial mathematical models on real conditions unless you have a standard mathematically defined distrubution

can you give an example please (link will be enough).

to me distribution is important on a calculation level, as i can verify if my model will work in the specified domain

 

  • Almost all models on the financial data are based continious methods even the discrete models (discrete signal processing methods were originally created for analysing of sampled continious signals). Engle in 1990s told the truth and said "actually all economic data are discrete by it's nature, irregularly sapaced, and isnt distrubuted normal. so it cant be analised the way we did before." He developed a new model, useless in real market analysis. but beter than nothing. a few followed him.
  • normally all of us used resampled financial data, resampled to make it regularly spaced in time,  but with every resampling the data explains less of its nature. (the daily data explains almost nothing about the original process.) But if we use it irregularly, we should solve the random time problem. we cant for now, we can only analyse it and Engle did that


well, this is obvious. PA (t) is discrete function, and not continuously defined across  the time domain.
Mathematically, ignoring this fact is wrong and leads to error.
Trying to compensate this by any method introduces error. What can you expect if you are correcting random input? Only that you will lose opportunity to exploit any regularities.

 

  •  the other issue is the linear randomness, probably because of its nature, prices mostly produce completely random signals (linearly speaking). I can enter the linear randomness but only if it isnt much for the topic? (a bit mathematics comes)

Please go on.

 

And on the end, after all this, how can you conclude that a mathematician doesn't have an enormous advantage over non-mathematicians?

 

For the mathematicians out there,

If a buy 1 lot  USDJPY is opened at 100.00, and now one wants to open a hedge sell at 105.00 (so as the equity going forward remains constant)

What sell size should be opened? Any mathematical formula for this? Thanks.

 

Rather than a mathematician I believe that a statistician would be more appropriate for this.

 
tme:

For the mathematicians out there,

If a buy 1 lot  USDJPY is opened at 100.00, and now one wants to open a hedge sell at 105.00 (so as the equity going forward remains constant)

What sell size should be opened? Any mathematical formula for this? Thanks.

You are off-topic. Open your own topic if you wish.

 
graziani:

However, describing itself and specificating mathematic methods doesn't make you a mathematician. you need to know and truly understand what you are doing.

It seems that you failed at the very beginning, by not understanding the consequences of the fact that PA is non-stationary stochastic process, which can be concluded out of claims you made.

after this comment and when you said "let the game begins", I felt an arrogancy there and behave offensive not personal, if you think it is a game then you should have expected an offensive player:

I never claimed I am a mathematician. but no one needs to be a mathematician to understand what they are doing. Einstein discovers the relativity by his feelings not with mathematics, then he used a simple transform to show it. He owes the mathematical background of the theory (which came after) to his wife.

So i don't know what you have tried to achieve, but it is meaningless to claim that it cannot be described mathematically.

And on the end, after all this, how can you conclude that a mathematician doesn't have an enormous advantage over non-mathematicians?

Did I ever said something like that, I said and always say in an obvious way: no one needs advanced mathematics here, intermediate statistics at most! 

statistic is a part of mathematics and consists a mathematical background, but we dont need advanced stuff here. so please dont manuplate my comments or read them before answer.

But if you insist I say now: Knowledge is everything in market, but it means any kind of knowledge about market. so being a good mathematician not makes you better than being a good Psychologist, here


Point is that it is a random process. Function of a random variable is a random variable

 if it is the randommness eventually, why do you insist on stationarity. stationary is solvable but randomness absolute. so again why do you insist on stationarity.

Actually i believe that we agree about this.

 Actually, I might have been happy for that but then you continiued:

CLT is mathematically formed. And it is proved in more then one way. So it is correct.
Mathematics is very exact. To prove that a theorem is false, you just need to specify one case for which it isn't true. I didn't see such prove.

So this is a matter of false interpretation. What does CLT say?
It says "bla bla bla bla  when n -> infinity bla bla bla".

So the CLT doesn't claim that you will have normal distribution on any sub-domain in time (ok, domain defined by time period), but that it will have normal distribution when number of samples goes towards infinity.  

 No one said CLT is false, (again dont manuplate my comments): I said: Economic data doesnt conform the CLT axioms. and complained about: none cares the axioms. CLT doesnt tell every large population must be distributed normal. 

PA data is not normal even in so large data samples. please at least research about Mandelbrot hypothesis or heavy-tail distrubutions. you dont want see the facts.

  • why  is a distribution is so important: because there is no standard way to proof financial mathematical models on real conditions unless you have a standard mathematically defined distrubution

can you give an example please (link will be enough).

 

almost every paper involves emprical proof is an example of that, do you know to use google?, it is the basic emprical proofing process. every scientist wants to a generalised standard, because it is more simple to apply a standard way to solve many problems, than aplying a specified way for a specific promlem. it is the basic way of science: "trying to standardize and to simplify everything". if you try to argue it find some one else... not me.

as i can verify if my model will work in the specified domain

 even so, many researchers (and many genius) tried to do that, so far no specific aproach could be prooven, (or like Engle's example useless in real conditions), so if you can develop one, develop... it would be funny to see.

and for linear randomness if it is not off-topic? (this question is for moderator) I love to 

 
Sometimes good Mathematicians trying to figure out how to make money in the market reminds me of nerdy guys in high-school trying to figure out how to pick up women. (I was one of them)
 

There is something very important missing here.

There are too different Mathematicians, one  of them are theoretical Mathematician and the other is applied Mathematician.

As you know theoretical Mathematician are people those discover new theory and new equations, and so on.

Applied Mathematician are more like data miner, quantitative traders, applied statistician, etc.  Even professional gambler, who uses Mathematics to make their gambling decision will be on this applied Mathematician category. 

Most of time, theoretical Mathematician won't be able to make money in forex market. They become famous because they wrote some interesting articles or journals and not necessarily their theory has any predictive power. 

There are many example, like Black-Sholes, Nobel prize Winners who gone bankrupt with their Black Sholes theory.

Applied Mathematician can make money in Forex market, if they are well trained in their own discipline.

Theoretical Mathematician and Applied Mathematician are like Ice and Fire. You can't put them in one glass as they don't agree with each other.

Kind regards.

1234
New comment
Reason: