Walk forward analysis

Hi everyone,

 

Whilst contemplating the while strategy selection process, a thought popped into my head.  Excuse the question if it is idiotic, it’s been a long day today.

 

When doing WFA, there are X parameters I can vary, and of course each one has a start, end and step.  Surely the outcome of WFA is going to be very dependant on which of the X parameters I chose to vary, and the range and resolution I select for each one?

 

For example, what if I chose not to vary one or more parameters, or chose a very wide range or narrow range of values for the parameters?

 

How can I be sure the decisions I made when setting up the WFA are correct?

 

Cheers,

 

Mike

I always know what SL/PT/TS/TP I want before WFA so I never optimise these values.  I only change entry parameters.  I normally know which components of the entry parameters I want to remain static over the time for example “close(1) > MA(200)” so I wouldn’t alter these parameters.  Basically, I normally only alter between two to four parameters but this choice has to be driven (in my opinion) by your knoweldge of technical analysis etc.  In terms of parameter range I always select the 300% because if the model can’t handle a large parameter range then I am not interested in trading it.

Do you have any live posetive trading reults showing the results of using the settings derived from this type  WFA in comparison with the original strategy (portfolio). ? I have some but it to soon to say since it only been a couple of months going live but the WFA portfolio have had a smoother equity curve but similar total performance.

Pardos chapters on optimization and WFA will help with the theory behind all this. Specifically page 217 explains how to choose parameters for optimization.

Sent from my iPhone using Tapatalk

@notch: Can you please clarify? Where do set this setting: “In terms of parameter range I always select the 300%” ?