Please try to understand the following:
The forecast array has 12 columns. For the sake of argument, if each column had a turning point, what would the predictive value of it?
Zero.
If on the other hand, a forecast array had only a single turning point in it, then that would be something. But it would still be of a low value because it can be interpreted as both a low or a high.
The fact is that forecast arrays typically have more than one turning point, in fact many turning points, so the predictive value is low due to the resulting ambiguity.
Do we have any idea how he actually creates the arrays?