Methodology

9/3/20 Multi Strategies, Proving a Point

The prior post showed new code for the hex build. I might have made the code copyable if I knew how, but it is probably better to enter it manually, or even better, figure it out yourself if you want to investigate this.

For ColXSel, the 10 – 17 pattern is given and the VBA routine pairs 10 with 14, 11 with 15, 12 with 16…etc. That gives us the hex codes for xEM.

The technical problem is how to manipulate the array to do different logical combinations.

There are several other indicators I look at other than E (EMA) and M (SMA). These include P (Percent of Range) and R (Natural Log of Rate of Change).

The table shows what I’m trying to explain. The strategies in the first column should be clear and then the numbers are aggregated and don’t have much significance except higher is better if the idea is to accumulate money. To get average numbers the entries can be divided by 8 but that has meaning mostly for days.

Note that m13m21 performs better than em13. P looks not too interesting, I might replace that with S (Linear regression slope).

R does pretty well. When I started the project last year, I was looking at R combinations.

The point is, assuming everything is kosher, the numbers for virtually all the lengths and indicators are sensational. This cannot be said for strategies developed from consecutive time. The median strategy listed here will blow away any consecutive time strategy by probably over a factor of 10.

Leave a Reply