Everybody agrees
that a highly optimized system runs the risk of curve-fitting, so one should
optimize on only part of one's data, then backtest the optimized system on the
rest to see how it holds up.
Is a feature to do
this in the works, or does anybody have scripts that will help me automate this?
The last thing I want to do, after I have optimized a system with a few thousand
parameter combinations, is backtest more than a couple of those. The kind of
thing I have in mind is a script that would optimize a system across, say, the
earliest 50% or 2/3 of the data in some date range, then backtest each
combination across the latter part. If there is already such a script, so I
don't have to code it, myself, that would be great.
Thanks,
Ed