I'm struck into an portfolio opt. problem and the paper I'm replicating (or, better, trying to) is using a "Standard Grid Search".
Since I never encountered it before, I would like to ask you about: what's the intuition behind this numerical method? What's it used for? How could it be helpful for optimization problem?
Please, share with me some insights!
If you have any link (no wikipedia, already tried) or material, it's very welcome!
Thank you in advance!!
p.s. I understand the purpose of this numerical method is to avoid numerical convergence problems and local optima issues. But I would like to have a longer explanation (or at least less cryptic).
The grid search is, in essence, systematically search through all possible (hyper)parameters to find the best one.
So, for example, if your portfolio depends on two hyperparameters $A,B$ (say taking values in $[0,1]$), then you might search through the $1001^2=1\,002\,001$ possible pairs $$ (A,B)=(0,0), (0,0.001), (0,0.002),\dots,(0,1),(0.001,0),(0,001,0.001),\dots,(1,1) $$ to find the optimum $A,B$ to 3 decimal places.
This method is easy to implement, but is very inefficient when the number of hyperparameters grows or you want higher precision. However, this is the only method that will guarantee you the absolute optimum.