Projections research

For the 24h projections to be of any use you need them to land in the same session cycle as the actual high or low, or if there is a change in the dominant cycle, then the change in 24h cycle translation (and therefore also the projection accuracy) should be explained by this change.

This morning I started working on improving this projection accuracy. I thought about writing an indicator in order to extract and forecast the cycles, but it would take too long. Therefore I decided to do it on paper first. Here is a screen shot of the first draft:

There are some quite big differences, so I am going to go back and revisit the data and check the accuracy, and also see if I can come up with a more consistent set of rules.