Why is the normal distribution not normal? - page 45
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
As early as page 29 I was surprised at such support for the cost flub I started, it's already page 44...
Of course it's all for a reason. The passage (see attachment) describes a phenomenon closely related to the observed one.
// Excerpt from "Don't growl at the dog" by Karen Pryor
Justice you say... Heh...
Let's say you and I are friends, and I say listen friend, we got a package of 10 oranges each, 8 of them.
And you're like, "How's that?"
.
I said: I don't mind, I ate my 8. :о) Is that fair?
Fairness is when the maximum number of bids can get the maximum chance of selling, no one is denying the fact that over time, bids move (those are removed and replaced by the same volumes at a different price) price moves to meet bids and bids move to have a chance to be filled.
If you have a pending order at who knows where in the market, what are the chances it will be filled, and what are the chances of the next order?
You have a very subtle sense of justice. :о)
You have a very delicate sense of justice. :о)
Yes.
I don't have any. In this matter, I trust only robots...
;)
Yes.
I don't have any. I only trust robots in this matter...
;)
You shouldn't trust robots, but those who program a "sense of justice" in them. Otherwise you'll end up with no oranges at all.
At least I got two ....
:о)
We should not trust robots, but those who program a "sense of justice" into them. Otherwise you will be left without oranges altogether.
At least I got two ....
:о)
Fair enough.
;)
I am working on the distribution. The distribution of the size of the bars and their offsets relative to each other.
I correct the distribution with the sigmoid ((2.0/(1.0+MathPow(2.0,-In*(10/L))))-1.0)*H
where:
In - input value
L - "length" of the sigmoid
H - "height" of sigmoid
It turns out like this:
red histogram - distribution of bar sizes
black histogram - distribution of bar offsets
green line - distribution of modified bar sizes
blue line - distribution of modified bar sizes
What else can be done to "straighten" the distribution? Ideally, it should be a straight horizontal line.
PS sorry, code for 5. I am too tired to write it for MQL4. :)
I'm working on the distribution. By distributing the size of the bars and their offsets relative to each other.
I correct the distribution using sigmoid ((2.0/(1.0+MathPow(2.0,-In*(10/L))))-1.0)*H
What else can be done to "straighten" the distribution even more? Ideally a straight horizontal line should be obtained.
You are not actually "straightening the distribution" but scaling it, if you remove the black and red histograms from the separates, the blue and green lines will also be a bell.
Nope. I don't scale. The Y-axis is the number. For all four distributions, the total number of values is the same.
The X-axis is clusters (classes).
Look at the code, please.
Nope. I don't scale. The Y-axis is the number. For all four distributions, the total number of values is the same.
X-axis is clusters (classes).
Have a look at the code, please.
Yes, I've looked at the code now, everything is correct (I don't really like fives yet :o)
The shape of the distribution depends on the ratio of RMS to RMS. The bigger the ratio, the dumber the distribution bell.
At one time I did this straightening through the root of the bar, and because the curves of the chart constantly diverged, I wrote an indicator, where zero was taken every new day or week, it turned out something like an indicator of deviation from the averaged hover (a kind of indicator of return), but it often happens that does not return in general I decided that it has no future.