To reduce the risk of over optimisation, when re-optimising a model I normally allow the model’s parameters to vary by no more than 5% of the original value. How many percentage points do you allow your model’s parameters to change when re-optimisating? I wonder if I’m being too conservative and missing out on more potential profits.
it is hard to say in general, it would depend also on the type of parameter. Small change of some parameters might have bigger impact on strategy than other.
For example, if you optimize period, there is no big difference between 2 and 3 values, but for standard deviation difference between 2 and 3 is huge.
I personally use 10% or even more, bug again, depends on the parameter.
You must be logged in to reply to this topic.