From theory to practice - page 62
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
Nikolay, I had a quick look at OPEN/CLOSE minute bar allocations, the link to whichYuriy Asaulenko provided me.
Yes, it seems that we must work with these prices, or rather choose one of them. Here you have delta T = 60 sec. and the distribution that is close to the Laplace distribution.
Though, I will still look at it - there is no need to hurry. The moment is very crucial.
There are holes in the bars, choose Open, and we will fill the holes with Close of the previous bar.
I understand that calculations will be difficult for many symbols, especially because the open price is set once in real time and does not change anymore, while the close price dynamically changes every tick until the bar is fixed in the history.
The rows are identical with one tick shift. Generally, Open[i]-Close[i-1] are increments of one tick, measured with frequency of 1 minute.
There are holes in the bars, select Open and fill the holes with Close of the previous bar.
As far as I understand the calculations will be difficult, even more so for many symbols, the open price is set once in real time and does not change anymore, while the close price dynamically changes every tick until the bar is fixed in the history.
The rows are identical with one tick shift. Generally, Open[i]-Close[i-1] are increments of one tick, measured with frequency of 1 minute.
Yes, yes... Apparently, we have to work with such a price series...
Question - then why did you fight for ticks and their history???? А??? :)))) Probably thought it would be more accurate? And that's how it turned out. It is impossible to work with ticks at all - extremely high hardware requirements and there is no delta T at all... How do you solve the equations? :))))))
In addition, it is now that people from different DCs have a "language" to communicate - precisely at the level of minute OPEN/CLOSE. Do you agree?
In addition, we have a sample of 240 values for the H4 timeframe. This is a nice sample that you can and should work with. Isn't it?
I just checked, a sample of 225 values is enough to cover 95% of Laplace distribution. Well, it all fits!
And what do you mean by '95% distribution'?
95% of what?
95% of what?
Do you even know how the sample size is calculated?
I don't really care how you calculate it. I'm asking why you think a sample of 100 ticks and a sample of 95 ticks will have fundamentally different distributions.
You should write that as 95% of the sample size. Or else "on the distribution". There's water in the wording.
I don't give a damn how you calculate it. I'm asking why you think a sample of 100 ticks and a sample of 95 ticks will have fundamentally different distributions.
Did I say that? No, look - in order to talk about confidence levels, we have to know exactly that our sample size covers a particular distribution with a certain confidence probability. This is found from Chebyshev's inequality. That is, if we choose a sample size of 240 values, then we are sure that we have covered almost the entire Laplace distribution. And then exceeding the confidence intervals (or rather tolerance intervals) calculated from the quantile function will indeed tell us that some limit has been exceeded, beyond which the price will rise with less probability than it will fall.