Early on you described a strategy for using a WN History strategy. In section 1, you made a few statements.
Statement:
"That depends very much on the shape of the history differences table and on the estimates from the Sum Chart and Differences chart."
This is the statement that has raised my eyebrow. Of course, there are sentences before this one that probably sets the context.
Statement prior:
"However it's possible to use a wider interval, especially when expecting a higher negative difference of the future sum."
Soooo.... Talk to me about the shape of the history difference table and if that drives the estimates of the Sum and Differences Chart.
I have genuinely been practicing my skills at setting the ranges for the 11 different sum values. What upsets my ranges outside of being genuinely wrong in a given instance for a given direction. It is the unxpected large magnitude change in a sum (positive or negative) that is seen in the summary table.
What I have seen at least in my game is that the sum values do not oscilate from game to game up and down. I guess what I am saying is that when a sum value is moving in a certain direction there is no reason for you to believe it is going in the opposite direction (in general) until it does at least a few times.
Sorry if my thoughts seem very fragmented. I do see trends in the sum. Tonight is the first time I have set the sum filter values that have returned over 5000 combs. We will see tonight how well that does. What I do like is when I set my sum filter values that return very few values. Which is what has been happening to me most of the time. If I could some how understand the random large magnitude sum change or at least which column that will exhibit that behavior. I might be able to trinagulate ...
Anyways I am rambling a bit now. However in my random thoughts I was hoping to understand more about shapes in the history difference table ...