When creating a Building Block strategy, I would like to evaluate multiple indicators, and only enter a Position if a sufficient number of indicator conditions are true. Ideally, I would assign a numeric weight to each individual indicator, and a Position would only be entered if the sum of the weights are greater than some threshold.
Does Wealth Lab support this? I am new and still learning, I apologize if the solution is obvious.
Does Wealth Lab support this? I am new and still learning, I apologize if the solution is obvious.
Rename
QUOTE:
only enter a Position if a sufficient number of indicator conditions are true
All indicator conditions added to a single Block are ANDed together. Is the idea to "vote" for at least X conditions out of Y possible to be met on the current bar to make an entry? if so then there apparently is no direct way to accomplish this...
QUOTE:
a Position would only be entered if the sum of the weights are greater than some threshold
...however, the "N Times within N Bars" qualifier added to each condition block may serve as a workaround?
QUOTE:
Ideally, I would assign a numeric weight to each individual indicator, and a Position would only be entered if the sum of the weights are greater than some threshold.
What you're describing here (numerically) would be done with the Neural Network extension on WL. I don't think the WL parameter optimizers are suitable enough to solve for the weights (eigenvalues) you're looking for.
You could also install an external numerical analysis package like Math.Net, which can solve linear systems (eigenvalues and eigenvectors) for you. But I would very strongly recommend trying WL's Neural Network extension first unless you're a numerical analysis hack. In addition, neural networks will solve nonlinear (i.e. event driven) systems, which standard numerical techniques (single-value decomposition) cannot do.
If you're having trouble with the WL Neural Network extension, start a new topic and pose your question about it there. I "think" you would interface the Neural Network extension like any other indicator with blocks, but I'm only guessing here.
Your Response
Post
Edit Post
Login is required