Methodologies or how to use EWFO
The most ideal fitness generally the one that has good performance metrics and that gives the most stability between the out of sample and in sample results. In the example below the parameters and hence profit is identical since 2013. This means the optimal parameters have been found, and we hope will still be stable for future unseen data.
We also want to see stability in each parameter that has been chosen.
Note on this example, all parameters are stable, and are the same since run 8 onwards. This is excellent. If a parameter is erratic, it is best to investigate if the use of the parameter is valid.
For example I have seen a system builder produce coded as shown.
value1=average(range,x);
The parameter was much more stable when the following code was used.
value1=x;