optimization in Strategy tester

What is the purpose of optimization in strategy tester?

Can anyone please explain or point to a link that explains this in detail. Is it worth using in strategy tester? why or why not?

The point of optimization is to find out which combination of settings would return a required level of the optimization goals (which could be: greatest profit, lowest drawdown, etc.).

Note, however, that optimization can only tell you what would have been the best in the past, and you may or may not use this as a suggestion of what settings to use in the future.

Thanks,
Ravi

A general rule of thumb I use when optimizing is to use a method called "rolling optimization"


Example:

I have 10 years of tick data.

I take the first year--optimize heavily for that--and run it against the second year.

I take the second year--optimize heavily for that--and run it against the third year.

Repeat until you use up all 10 years, and pay attention to what universally works and what doesn't. Once you find those optimal parameters, run it again, against all 10 years.

If you did everything properly, you should see a uniform equity gain across the 10 years and then you know that the EA was optimized properly. (as I see it). Repeat as often as possible.

I do (nearly) the same thing on a week-by-week basis with my simple ma cross strategy. What I do is look at the optimized values for one week and use exactly those for the next week, ad infinitum.

Its not great, but it gives me at least a 1% increase per week when I risk 2% of my account per trade.

I would recommend using a larger test period. A good ratio would probably be 20:1.

You use the previous 20 periods to predict the 21st period.