Numerical series density - page 20

New comment
 
Vyacheslav Kornev:
I don't think it would be useful just to look for clusters. Better to calculate a landmark from the previous day, e.g. centre of mass. That's enough.

Enough for what?

 
-Aleks-:

You can't be so categorical - it hurts progress...

That's understandable - and I've already written you why... but we have to consider the relationship of each cube to its neighbour...

We can find it, but only if we know what we are looking for... which complicates the search - I have indicated the reason above.

About the example - theoretically, let's take the numbers and:

1. Let's increase the number series by the same number series, previously multiplying it by 1000

2. the same point as 1, but replace 56 with 59

I simply add a minimum value of 1 point in my algorithm.

Heh, so you're not going to consider them at all? Oh, okay.

I will tell you how. But I still don't know what you need. Have you read the example of the swing?
 
Vyacheslav Kornev:
Heh, so you're not going to take them into account at all? OK.
I'll tell you how. But I still don't know what you need. Did you read the swing example?

How is it that I'm not going to take them into account - they increase the density. Your method will be interesting - tell me.

About the swings - just this is proof that this approach - as finding the point closest on average to any other point - is questionable, which I have said more than once - and suggested a new calculation that should prove it - if I am not mistaken.

 
-Aleks-:

Enough for what?

What exactly is it for? Determine the price on the chart where something will happen, isn't it?
 
Vyacheslav Kornev:
What is it for? Determine the price on the chart where something will happen, isn't it?

Hmm, that wasn't the goal - the original idea is that people use different omens - tech analysis, but everyone's omens are different, and if many omens call about the same price, then that area will be significant to the market, as a large number of people expect the market to change and will be influenced by it.

If you set aside my idea, then using your proposed method it will be possible to find a certain point - it is not clear which window to take - in the script took 100 weekly candlesticks. I will not predict the result yet, but it will either be a narrow strip, or on the contrary - very wide, which is more probable. How useful it is, you can only find out by experience.


 
I can't do that. It's like I'm giving examples in the void. I tell you about Thomas, you tell me about Yourself.

OK. Please draw a series of numbers of any ok with any deltas. And BELOW, please, what you want to highlight on it.
 
Vyacheslav Kornev:
I can't do that. It's like I'm giving examples in the void. I'm talking about Thomas, you're talking about Thomas.

OK. Please draw a series of numbers of any ok with any deltas. And BELOW, please, what you want to highlight on it.

So here's an example https://www.mql5.com/ru/forum/163871/page6#comment_4083680 - the result is quite good.

 

It would be good to program your algorithm - it's hard to calculate everything in Excel...

 
Uh, well, you found 20 points. If there was a bigger row, you'd have found 100, 7,000.

What are you going to do about it?
 
Vyacheslav Kornev:
Well, you found 20 points. If you had a bigger row, you'd have found 100, 7000.
What will you do about it?

So you're asking me the same questions I'm looking for answers :)

That's why I'm wondering what criterion to use to sift out the clusters - and whether sifting them will be correct.

I can sift out more deltas - I regulate % of remaining deltas relative to all numbers - now it is 50%, but I can also sift out 10% - your algorithm does not allow me to automate this process.

Now according to your algorithm I have 132 different deltas - how to filter out the right ones is a question - it is clear that from the smallest one, but it is not clear to the largest one.

1...1314151617181920212223242526
New comment
Reason: