Optimizing timeframe?

Posted By: ibra

Optimizing timeframe? - 04/21/14 06:52

Hi,

I've never thought of it before, but then I looked into the thread (http://www.opserver.de/ubb7/ubbthreads.php?ubb=showflat&Number=431959#Post431959) that dusktrader made which "suggests" that one also could optimize the timeframe for a system.

I've read in the manual that this is supposed to be done in a special way.

Quote:

Parameters that affect the price data array, such as BarPeriod, require special handling for optimization. They can not be changed during the simulation, so they can not be walk forward optimized; for a walk forward test they must be optimized in a separate training run before all other parameters. They can also not be read from the .par file, so their optimized value must be entered directly in the script as start parameter of the optimize call before further optimizing, testing, or trading the strategy.

What's your though about optimizing the timeframe? Is there a reason NOT to do it?

The thing I come to think of is, that the most common timeframes being used in trading is 1 hour, 4 hour and the daily. It just feels weird to me if a systems selects like for example a barperiod of 2 hours and 35 mins... You know what I'm saying?

Ibra

Posted By: jcl

Re: Optimizing timeframe? - 04/21/14 08:34

I would not optimize the time frame. Better optimize time frame dependent parameters directly, such as the time period of indicators.

Posted By: dusktrader

Re: Optimizing timeframe? - 04/24/14 01:34

I'm on (ancient) mobile at the moment, so I can't reference the link you posted above... but I would just like to comment that I think the biggest danger in optimizing BarPeriod could be overfitting.

It wasn't until Zorro came along that I even had the ability to do something like that, so it was an interesting thing to try.

My thought is that it could be helpful to "zone in" on a good frequency of bars, assuming you have a rudimentary logic and want to see which bar frequencies that it may be relevant in.

So far in my testing the optimizer tends to pick standard frequencies anyway (ie 15min, not 14min for example). Presumably this is because of the bigger tides.

Neat feature tho - I'm glad we have it at our disposal.