Understanding the underlying mathematical idea

[white07S]

Considering the basic knowledge of support and resistance, I would like to introduce the underlying mathematical idea.

1. Finding global maxima and minima in a time series data
2. After obtaining the pivot points, choosing two of them would be simple because any pair of minima or maxima points would form a line, creating a total of n(n-1) complete lines, all of which are perfect fitting lines because basic geometry states that 2 distinct points can describe a line. However, there are n(n-1)(n-2) methods to choose 3 points, and this enumeration has an algorithmic complexity of O(n3). There will be some mistakes because these points are not always in a straight line. Since we must exclude wrong point sets and include correct ones, the error level will be crucial.
3. Few things to read:

• https://en.wikipedia.org/wiki/Numerical_differentiation
• [Library for Numerical differentiation] https://github.com/maroba/findiff
• [Alternative for line finding algorithm] https://en.wikipedia.org/wiki/Hough_transform

Drop down a comment in case of any doubt.

[lukik45]

@white07S Is that what you mean by pivot points? https://www.investopedia.com/terms/p/pivotpoint.asp ?

[lukik45]

@bujowskis I found a neat explanation of Hough transform https://www.youtube.com/watch?v=XRBc_xkZREg (essentially the first 11 minutes)

[bujowskis]

thanks @lukik45, didn't sit down to it yet but will definitely check it out

[white07S]

@white07S Is that what you mean by pivot points? https://www.investopedia.com/terms/p/pivotpoint.asp ?

Yeah that's exactly what I meant.

[bujowskis]

Okay, after a few days of getting the hang of it, I think it's to some degree clear. I think I may be missing something regarding the application of the Hough transform, but I can't really formulate what it is, so I think it's best if we keep going and if it arises, I'll ask you two

[bujowskis]

I think there's nothing else to add on here, so I'm closing the issue