FR H-Volatility - page 41

New comment
 
Prival:

Of top, but I don't know where to write. If I am not mistaken the Guinness Book of Records has a record of 1200% p.a. Larry Williams http://web-investor.academ.org/index.php?action=articles&id=71


We have some good traders here as well :)

http://www.rts.ru/main/investor/

guys 20-30 years old

No demos, only real trading, mostly scalping on the RTS Index futures, in annual terms the leader of the championship earned approx. 7000% p.a.

No.2 in the list is an options man.

I did not take part in the competition

I've been in the stock market myself for about 6 years, since april of this year on my own account

 
torgash:

We have some good traders too :)


And there are no OUR traders or yours here at all. All of us are native :-). Only I have an inaccuracy. It's not 1200%, it's 12000%. I wish he were ours :-)
 
Prival:
There are no Ours or Yours here. All of ours are native :-). Only I have an inaccuracy. It's not 1200%, it's 12000%. I wish he were ours :-)


It is clear that it is 12000, but that is not the point. The matter is that American traders are very skillful in self-promotion, while domestic traders are withdrawing money quietly. By the way, in his book "Long-Term Secrets" Larry admits that his system sometimes did not produce results even for a year or more. This is one.

I can easily tell OURS and NON-OURS by their mother tongues. That's two.

And you need to broaden your horizons, gentlemen. I mean, there are other markets besides FX, from which you can earn good money for a new tie. The bots at the link above have also traded and have shown quite decent results. I've already learned QPile and created bots for QUIK. I wish the same for everyone. This is three.

 

In one book I found the following explanation, after proving some mathematics:

"The bottom line is that with zero initial capital it is possible to construct a portfolio of securities such that the expected return can be as large as desired and the risk expressed by the variance of return as small as desired. This is called arbitrage: a positive return is achieved with zero risk and zero initial capital.

Very, very good explanation.
 

Hi NorthernWind.

In the article I linked to above, there is a description of the methodology for assessing the significance of the information used for forecasting. Here is an excerpt:

The dip method allows to quantitatively measure the predictability of real financial instruments, i.e. to test or refute the market efficiency hypothesis. According to the latter, the dispersion of points along all coordinates of the lag space is the same (if they are equally distributed independent random variables). On the contrary, chaotic dynamics providing a certain predictability should cause observations to cluster near some hypersurface, i.e. the experimental sample forms some set of dimensionality smaller than the dimensionality of the whole lag space. To measure dimensionality we can use the following intuitively understandable property: This fact is the basis for determining the dimensionality of W sets with the already familiar box-counting method. The dimensionality of a set of points is determined by the rate of increase of the number of boxes which contain all the points of the set.

Below is the figure with the dimensional(informative) increments of this set, calculated by the box-counting method for S&P500.

and it is argued that in a 15-dimensional dip space the experimental points form a set of dimensionality of about 4. This is significantly less than the 15 we would get based on the efficient market hypothesis, which considers the incremental series to be independent random variables.

I have a misunderstanding relating to:

1.It turns out that the information dimension W is equal to the number of inputs D only for the case of a uniformly filled NE of the whole phase space. Already, for CB distributed by Gaussian, for D=1 - W=0. 7 i.e. series of increments in this case is not random!

2. For realization of method, when "dimension of set of points is determined by the rate of increasing number of cells (boxes), containing all points of the set", we should take by one or another way n-fold integral (sum), where n is certainly more than 10-15. This is not realistic!

I don't get it...

Colleagues, what do you have to say about it?

P.S. The trick of box-counting method is that it takes into account possible non-linear relationship between inputs and outputs (as opposed to static methods). This, in turn, is important to determine the optimal number and quality of inputs for a neural network.

 

Hello, Neutron! Happy holidays past and upcoming.

I apologize of course, but this box-counting method reminded me of the books about the miracle method of urine therapy. I am not a supporter of dynamic chaos, fractals and the like.
 

Neutron

Thanks for reminding me about this stuff. (I missed it, just looked it up).

Finally, more or less sensible info on NS.

Here's what the output should be (at least in the simplest case). "A network with an output nonlinearity of the form ... is trained to predict the sign of the change and outputs a sign prediction with an amplitude proportional to its probability. " (p.12).

Not Buy and Shell.

I've said this many times before. A lot of people don't believe it. Neutron sorry to be in this thread and repeating myself just want to warn others against this mistake. A lot of people do not think about it and make the exit of NS Buy and Shell. I do not know where it came from (I am afraid it came from Reshetov again). But you should never step on this rake (Buy and Shell). It's a very fine point, but think about who has been working on the NS for many months and still can not get a decent result. Maybe I am right + the authors of the material posted Neutron + Batter, and not Reshetov. We cannot have Buy and Shell as the output of NS.

Here is where Better says the same thing.

https://www.mql5.com/ru/users/Better

https://www.mql5.com/ru/users/Better

Predicting the sign (direction, velocity vector) of the rate and predicting Buy and Shell, are completely different things. You cannot forecast (predict, recognise) what is not on that curve.

 

Neutron

Thanks for the archive of minutes and ticks. Need some advice on the contents.

There are two files.

The first one is EURUSD 1m.prn.

Here are some lines from it

20070801,000000,1.3679,1.3679,1.3678,1.3679,9

20070801,000100,1.3678,1.3679,1.3678,1.3679,4

Going in order, the first is the date, the second I don't know, then Open, High, Low, Close, Volume.

Please, give me more information about the second date and if I have made a mistake, please correct it.

The second file is EURUSDtick.prn.

Here are some lines from it

2007.08.01 00:00 1185926430 1.3678 1.368

2007.08.01 00:00 1185926447 1.3679 1.3681

2007.08.01 00:00 1185926448 1.3678 1.368

2007.08.01 00:00 1185926451 1.3679 1.3681

2007.08.01 00:00 1185926452 1.3678 1.368

2007.08.01 00:00 1185926455 1.3679 1.3681

2007.08.01 00:01 1185926461 1.3678 1.368

2007.08.01 00:01 1185926472 1.3678 1.368

2007.08.01 00:01 1185926472 1.3679 1.3681

2007.08.01 00:01 1185926488 1.3678 1.368

2007.08.01 00:01 1185926490 1.3679 1.3681

2007.08.01 00:01 1185926508 1.3678 1.368

2007.08.01 00:01 1185926509 1.3679 1.3681

2007.08.01 00:02 1185926562 1.3677 1.3679

First number is date, then hours:minutes, don't know, Bid, Ask.

Enlighten on the third number, what is it with what is eaten.

And 1 more question, I compare two files, at zero minute volume=9, and ticks (second file) like 6, then at first minute volume=4, and ticks 7. I must be mistaken somewhere, something I do not understand.

 
Prival:

The first one is EURUSD 1m.prn I understand the minutes.

Here are some lines from it

20070801,000000,1.3679,1.3679,1.3678,1.3679,9

20070801,000100,1.3678,1.3679,1.3678,1.3679,4

Going in order, the first is the date, the second I don't know, then Open, High, Low, Close, Volume.

Enlighten me about the second number and if I made a mistake, please correct it.


PMMSS - i.e. time ;)
 

Yeah exactly PMMSS, the first file seems to be clear. But what about the second one and the mismatch of ticks and volume ?

1...343536373839404142
New comment
Reason: