Using Neural Networks in Trading. - page 14
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
That's weird, it's random ((
Why is it random?
Here's a quote from Pastukhov's thesis:
As you can see, this value can be considered a constant.
)) I'll cite a stronger authority for me.
'FR H-volatility' because you never found H. It jumps. It's a random value if you do it the way you researched it. Although I may not have understood everything there (I have not seen all studies).
Pastukhov should have defended his dissertation, this in the first place. You were looking for a pattern = an opportunity to earn I have more faith in your research. But h is, it is equal to twice the value of the spread, at least for the pair GBPUSD, more precisely in terms of radiolocation it is the value of movement that can be detected. This value is related to the threshold signal/noise ratio (I got it around 1.6-1.8 of the spread).
I'm interested in how the championship organizers arrived at this value.
'FR H-volatility' because you never found H. It jumps. It's random.
You're right.
I'd already forgotten about the details - the neuronics in the MTS track H, lets its biological counterpart relax:-)
You make a good point, but you have no idea how easy it looks when implemented in MT. There is no need to create an indicator for this, everything can be put directly into the Expert Advisor. Here, look at my last post. It's almost a blueprint for Kagi partitioning (vertices are shown, but no problem going to the partitioning itself).
Thanks for the indie. I now need to improve my mql4 knowledge to move forward.
Yes - I have the same thing.
I need to ask Prival how to get the desired distribution (rectangular) from an arbitrary distribution in analytical form.
Here I have found it. I deleted it at home and left it at the forum.
>> Tics: amplitude and delay distributions.
Well, you, Prival, are quick!
Thank you. Let's have a read.
Well, you, Prival, are quick!
Thank you. We'll read it.
)))) super-operative...
But my point is different... In different parts of the forum I meet statements, say, on the X-axis we have a random variable (time)... here, for example, Prival writes:
There are so many ways to analyze a random variable in Y axis (MA, RSI, etc.) but we forgot about X axis while it is a random variable and we should handle it carefully and correctly.
a tick is not a random value! a tick - the price increment is really a random value in time, if you take time as the number of seconds elapsed since 00:00 1 January 1970... the graph of tick increments in time would be a Markov chain with continuous time, i.e. time moments there are random - by the way, that's why it's a waste of time... Markov chains with continuous time are much harder to analyse than discrete time)
and the transaction price is not a random variable, it exists at any time (theoretically)... In practice, of course, we may encounter such nasty things as timeouts or simple requotes... But it does not change the essence of the matter - the price of a transaction exists at any time, i.e. the function of price versus time is a continuous value (regardless of gaps)!
The only difficulty here is the weekend. If simple holes in the data can be patched, then weekends cannot be "patched"... But in principle, it's not a problem either... You just need to number the bars backwards - the last one will be zero and zero will be last... ...and you'll get a non-random value along the X-axis! That's it, any of your indices designed to study a non-continuous function will work correctly!
...
Here's a special search for you: http: //offline.computerra.ru/2003/512/29647/
Two prerequisites
price - continuous
Time is continuous.
DC digitises a continuous value. in any digitisation there are errors. quantisation noise (level error) +- pips and discretisation noise (time error). So by taking minutes of their cloze we kind of a priori agree that there was a tick at this exact moment in time. But in fact it's not so. I would say it's very clear on minute ticks (who works with them). For those who work with the clock it doesn't show so much.
That's why for the GMT spectra it can be seen as "minute" noise, and it can be explained by the sampling noise.
price - continuous
time - continuous
both statements are incorrect!
we sort of agree a priori that there was a tick at that exact moment in time...
no, we don't agree that there was a tick... we're agreeing that there was a price... these are different things...
And sampling noise is too much... >> making allowances for the fact that our "device" (i.e. the terminal) has timing errors - Sergei, that's overkill...