algorithm to group prices

[fxshrat]

May 4, 2016 20:47:01 GMT bobpit said:

aha! I had to click on the thumb-up to see the code.

Yeah well, if I do the thinking process for you then I think a little bit of clicking work is not too much to ask for.

[fxshrat]

Now that you have a code example may I ask what application you actually need it for?

[bobpit]

Hi

Thank you very much for the code. I never worked with matrixes so I am still trying to understand how you do it.

The application I was thinking is this:
I want to find the support/resistance levels. So I take the prices of the swing high/lows. Some of the prices are very close together, these form a stronger support/resistance level. I need to identify those prices and separate them from the rest.

[bobpit]

Hi fxshrat

I learned a lot about matrixes from your code.

I see that the core grouping criteria is this command:
if( int(val) == i )


However this leads to incorrect results if points with different integer values need to be grouped together. For example 1.99 and 2.01.


Consider this:
{{0.85, 0.86, 0.89, 0.95, 1.98, 1.99, 2.01, 3.39, 3.4, 3.43, 3.46, 8.4, 15.2, 22.1, 22.14, 22.4, 22.435}}

rows: 1, columns: 17

{
{ 0.85, 0.86, 0.89, 0.95, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 1.98, 1.99, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 2.01, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 3.39, 3.4, 3.43, 3.46, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 8.4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 15.2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 22.1, 22.14, 22.4, 22.435, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
}

The output should be:
{
{ 0.85, 0.86, 0.89, 0.95, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 1.98, 1.99, 2.01, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 3.39, 3.4, 3.43, 3.46, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 8.4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 15.2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 },
{ 22.1, 22.14, 22.4, 22.435, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
}


Anyway, I am not looking for somebody to write the code for me. I asked this in case there is already an easy solution to it.

Thank you very much for the time and effort.

[fxshrat]

May 5, 2016 9:43:59 GMT bobpit said:

However this leads to incorrect results if points with different integer values need to be grouped together. For example 1.99 and 2.01.

Your previous examples all made it look like you were looking for sorting by equal integer sizes.

But your second example is doable also.
You just have to alter my uploaded code a little bit.

See



Do you figure it out yourself or do you require the new upload?

[bobpit]

Sorry for the misunderstanding. My OP stated:
"I am trying to create an algorithm that will group price values, when they are close together."

If you went through the trouble then I am very interested to see what you wrote.

I could change your code slightly and use ROUND to achieve this output.  But this certainly is not a valid solution for what I asked.

[fxshrat]

No, no rounding. Why rounding?

You would obviously compare value with previous one(s).


As aside I've found a flaw in my 2nd method if original values are trending within the set threshold being used to compare one value with another.
Then those ones will be grouped together in a single row i.e if trending within the set fractional threshold of 0.3 between each value from 0.5, 0.7,... to ..., 9.6, 9.8, 10.1, ...

[bobpit]

yes, the threshold needs to be dynamic and adjust to the values set.

If the support/resistance lines are for EUR/USD, then the prices will be in the scale of 1.2372, 1.2373, 1.2375 etc. This is the toughest part.