Was programming, and came across an interesting problem thought id share it to see if anyone here has any insights on how they would approach this particular logic problem.
The Problem: Classifying “Highs” and “Lows” accurately w.r.t. scale of recent move. With a high accuracy rate (upwards of 95%)
Example of expected Classification: [https://www.tradingview.com/x/PlJvHlJV/](https://www.tradingview.com/x/PlJvHlJV/)
Intuitively it is very simple to classify Highs and Lows just by looking at the chart. But when trying to code that logic for this specific problem I noticed you encounter some interesting issues.
Note: (I’ve already found a workable solution for this problem but was interested to see how other people tackle this problem if they’re interested). My solution involves using Gradients of slopes