Optimization Problems

Report any bugs with the Tickstory Lite software here. Please give as much detail as possible so the issue can be identified and resolved as soon as possible.
Post Reply
HarriMiles
Posts: 2
Joined: Sun Oct 25, 2015 4:12 pm

Optimization Problems

Post by HarriMiles »

I'm having inconsistencies with my optimization results. If I input the parameters from one of the optimization sets, the result I get is different to the one produced by the optimization. However, this only happens with some of the optimization sets. Is there a fix for this?
tickstory
Posts: 5289
Joined: Sun Jan 06, 2013 12:27 am

Re: Optimization Problems

Post by tickstory »

Hi HarriMiles,

The only reason we know of that could cause this is the 4GB FXT limitation that the Strategy Tester has. Make sure your FXT are under this size by breaking your back-tests up into smaller periods such as 1 year. Some other threads which may help are:

http://tickstory.com/forum/viewtopic.php?f=4&t=950
http://tickstory.com/forum/viewtopic.php?f=2&t=967

Hope this helps.
HarriMiles
Posts: 2
Joined: Sun Oct 25, 2015 4:12 pm

Re: Optimization Problems

Post by HarriMiles »

I have found that I only get inconsistencies with sets that don't trade on a day of the week (eg. Mondays, Tuesday). I was optimizing the days that the EA should trade on along with other factors. I discovered that if a set from the optimization traded on all days of the week it would produce the same results when inputted the parameters. But when I tested a set that didn't trade on a day of a week it produced different results from the optimization when I inputted the parameters. This is most likely a problem with the EA's code not Tickstory. So problem solved I guess.
tickstory
Posts: 5289
Joined: Sun Jan 06, 2013 12:27 am

Re: Optimization Problems

Post by tickstory »

Hi HarriMiles,

Appreciate you letting us know. It's always a bit tricky to identify what is causing the differences with the EA which is why it's always good to start with a basic EA to ensure your setup is correct. After that, narrow down your optimization variables to so you don't have too many result sets to perform analysis over.

Regards.
Post Reply