Topic: Strategy testing three issues

Hi i was back tasting some strategies, i noticed there are three observations that could be a problem
1)  there is a differences in data presentation, between FST and FSB why is this happening, is this a glitch? The data was dl'ed straight from the MQ4 USDCAD 5M page
2) for some reason the FST would not take a trade in life trading yet the FSB took several trades at the same time.  ( i am back testing  seven strategies at the same time  but only one USDCAD ( this one) ( some of the other strategies are trading). The FST shows that it is OK to take a trade.
3) this strategy looks for two bars in same direction and takes position in that direction. The stop for long position is suppose to be close below MA 3 based on LOW's ... this works good for long positions however the for short positions the FST logic will try to close position when the price closes above the MA 3 also based on LOW of the bar, what it should have done is to base the exit on price closing above the MA3 HIGHS...  I not a programmer, but i will attempt to modify the MA indicator to reflect that option, I likely will need some help from your guys if that's OK  smile

Post's attachments

06-17-11FSB Graph .jpg 523.77 kb, 1 downloads since 2011-06-17 

You don't have the permssions to download the attachments of this post.

Re: Strategy testing three issues

http://s2.postimage.org/yqzt5gtg/06_17_11_MQ4_Graph.jpg

http://s2.postimage.org/yr9qdjs4/06_17_11_FST_Graph.jpg

http://s2.postimage.org/yrom7o84/06_17_11_FSB_d_Graph.jpg

Re: Strategy testing three issues

Comment your screenshots, where's the misinterpretation between FSB and FST? Was there a trade already running when it didn't open the trade? Have you tested your custom indicator properly?

The closing problem is not actually a problem, this just how FSB/FST works, you could sort it out by using logical groups I think. If you plan to go ahead with a customization, then take a BB for example, throw out the deviation part, introduce base prices for both lines and it's done!

Re: Strategy testing three issues

The data graphs of the FST and FSB looks different. It seems there is difference in timing for the bar closing/ opening probably in seconds.  The FST graph looks identical to MT4 but the FSB differs. I will merge the pictures to one so you can see the difference.    Yes i was running other robots at the same time, i don't think there were any other orders send in at the same time, neither does the log reflect any orders send or rejected. Re my ability to properly test a custom ind. That's questionable, as this is really my first attempt of creating a program. I used the " Three Bars Swing ind." As the base, followed it's logic and altered the algorithm to my needs.... it's just a basic strategy nothing to write home about lol ...  the thought was that if the FSB Can successfully use it than the FST should too.  Am I wrong in that assumption?

The composite has three graphs the first is FST, it is identical to the one under the  MT4.  Yet the FSB graph is different. Can you see what i mean ?  (ps sorry for the crudeness of the composite  I didn't spend much time on it wink)
Thanks for the heads up on BB i will try to adopt it  if it works i can post it smile

http://s2.postimage.org/zw0o2dt0/06_17_11_FST_MT4_FSB_Graph.jpg

Re: Strategy testing three issues

It seams dr.B has Market -> Fill in Data Gaps option ON. There are no gaps between bars on the FSB chart. That makes the charts of FSB and FST different.
You have to switch off that option.

Re: Strategy testing three issues

footon wrote:

the thought was that if the FSB Can successfully use it than the FST should too.  Am I wrong in that assumption?

Absolutely, some bugs can be confirmed only while trading live!
Here's an example:
http://s2.postimage.org/13i98n9t0/1_0.jpg

Nice strategy, isn't it, but when trading it live, mysterious losses appear, so when dealing with custom indicators, FSB=FST isn't always true. You have to confirm that backtesting results are valid!

Thanks  "One Man Army" for your input.  I should have been more clear about  when i  compare the  FSB and FST, I understand that the results may vary (slightly wink lol ), what i meant by the term : 'successfully' is that it had no reference to the actual performance but rather to the ability of both programs  to respond to signals generated by the algorithm of the indicator  on equal basses.  I will restart the computer and test this indicator solo as the markets open... it could just be something on my side.

Re: Strategy testing three issues

Popov wrote:

It seams dr.B has Market -> Fill in Data Gaps option ON. There are no gaps between bars on the FSB chart. That makes the charts of FSB and FST different.
You have to switch off that option.

Yes that was it, the data presentation difference is gone

thank You

Re: Strategy testing three issues

Glad that data issue is resolved. About the custom indis - if indi is coded right, FSB and FST will handle the indi similarly, when comparing trading results, FSB must match FST. If there's a big difference, something is not right with your indi.

Re: Strategy testing three issues

footon wrote:

Glad that data issue is resolved. About the custom indis - if indi is coded right, FSB and FST will handle the indi similarly, when comparing trading results, FSB must match FST. If there's a big difference, something is not right with your indi.

I understand that.  the way this has behaved so far  is that the FSB shows orders  in the same time the FST failed to issue the orders yet the info bar on it's side reflecting what the indicator says shows that short or long was ok.  the FST did not issue an order.  if you look above on the image  ( second post second image) all is a go yet no short order was issued....  the only thing that could have influenced this behaviour was that i had  eight or nine different FST running at same time... none of them were trading USDCAD... i will test it again when market opens

Re: Strategy testing three issues

One more tip - make sure the go ahead is given at the right time, if there's a bug in indicator it might repaint the signal!