Aren't "Within the Past N Bars" and "N Times within N Bars" doing the same thing essentially?
Right now you can apply one qualifier to one condition at a time. If you could apply a qualifier to a group of conditions, I think that would do the trick. For example, drop the qualifier on a "Buy at" block to apply to the combination of conditions.
I gave a simple example on the original post. Perhaps there is an easy way to accomplish this using the current blocks?
Example 1 Building Block [At least 2 are true] voting system
- Condition 1 (RSI2 < 50)
- Condition 2 (IBR < 0.2)
- Condition 3 (Close[0] < Lowest(Close, 5)[1])
In this instance I might want to optimize on how many are true/false
Example 2 Building Block [Select 1] iteration of indicators
- Condition 1 (RSI2 < 50)
- Condition 2 (Stoch < 20)
- Condition 3 (Keltner Position < 50)
In this instance I want to optimize from 1 to 3. It iterates through an indicator list. It's a way to optimize which oscillator works best. Ideally, it would be nice to be able to save predefined groups of these for quick analysis of strategies (i.e. one populated with a list of my favorite oscillators and their parameters).
The ensemble method feature request is similar to this but instead of simple conditions in the example above make them simple strategies. The thought process of this approach is that each individual strategy captures some incite/edge into the market and by combining them you can get a much better overall system. It's generally used as a voting system like example 1 but can also be used as an anti-system (uses short strategies signals to trigger an exit).
Example 1 Building Block [At least 2 are true] voting system
- Condition 1 (RSI2 < 50)
- Condition 2 (IBR < 0.2)
- Condition 3 (Close[0] < Lowest(Close, 5)[1])
In this instance I might want to optimize on how many are true/false
Example 2 Building Block [Select 1] iteration of indicators
- Condition 1 (RSI2 < 50)
- Condition 2 (Stoch < 20)
- Condition 3 (Keltner Position < 50)
In this instance I want to optimize from 1 to 3. It iterates through an indicator list. It's a way to optimize which oscillator works best. Ideally, it would be nice to be able to save predefined groups of these for quick analysis of strategies (i.e. one populated with a list of my favorite oscillators and their parameters).
The ensemble method feature request is similar to this but instead of simple conditions in the example above make them simple strategies. The thought process of this approach is that each individual strategy captures some incite/edge into the market and by combining them you can get a much better overall system. It's generally used as a voting system like example 1 but can also be used as an anti-system (uses short strategies signals to trigger an exit).
This is basically a request for a feature we already had in WL6, it operates the way you describe, allowing you to specify how many of the conditions need be true. Not sure if that value was optimizable in WL6 but it would be in WL7.
Looking more closely, you're asking for two different things.
The first is the Multi-Condition Group ala WL6 - "N out of these 3 conditions needs to be true." You could optimize N from 1 to 3.
The second is a new concept of a Condition Selector - "The specific Condition X out of these 3 Conditions must be true." You could optimize X from 1 to 3.
Perhaps the behavior of the Group (Multi or Selector) could be exposed as another parameter of the Multi-Condition Group.
The first is the Multi-Condition Group ala WL6 - "N out of these 3 conditions needs to be true." You could optimize N from 1 to 3.
The second is a new concept of a Condition Selector - "The specific Condition X out of these 3 Conditions must be true." You could optimize X from 1 to 3.
Perhaps the behavior of the Group (Multi or Selector) could be exposed as another parameter of the Multi-Condition Group.
Your Response
Post
Edit Post
Login is required