• Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Topic: EURYSD D1 OBOS MA Osc with Break Even

EDIT: The backtest of this strategy is not correct. Please read next posts.

During testing new Break Even function of FSB I found very promising strategy.
It uses only one indicator - OBOS MA Oscillator and a Break Even exit.

I normalized parameters (set to 50) in order to check for over optimization and also ran Over Optimization Report. Everything looks normal for now.

Here is the Over Optimization Report Chart:

http://s3.postimage.org/1o41ua2x0/08_08_2011_11_34_41.jpg

Strategy Slots and Balance Chart

http://s3.postimage.org/1o4a3ytdw/08_08_2011_11_37_28.jpg

Indicator Chart

http://s3.postimage.org/1o4idnjus/08_08_2011_11_36_39.jpg

The strategy file is attached below.

Post's attachments

EURYSD D1 OBOS MA Osc with BreakEven.xml 2.55 kb, 79 downloads since 2011-08-08 

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

Popov's Website

2 Reply by SpiderMan (edited by SpiderMan 2011-08-08 11:18:53)

  • From: Australia
  • Registered: 2011-07-01
  • Posts: 431

Re: EURYSD D1 OBOS MA Osc with Break Even

If you trade a percentage of account rather than a single lot the results are much better!


http://img59.imageshack.us/img59/9691/s … 028001.png


http://img59.imageshack.us/img59/9691/snap20110808200028001.th.png

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Re: EURYSD D1 OBOS MA Osc with Break Even

This strategy can be improved if we trade it manually.

The main idea is that we don't need to close long position at bar close if we have to open long again at that time.

It's better to check the signal first and if the new trade has to be in the same direction, the exit order has to be canceled. In that way we can save one spread for these days.

Popov's Website

  • From: Australia
  • Registered: 2011-07-01
  • Posts: 431

Re: EURYSD D1 OBOS MA Osc with Break Even

You could always get the indicator to calculate it's level using the price just prior to the close rather than the previous close. Then add the inverse condition as a closing condition. smile

  • Registered: 2011-07-26
  • Posts: 31

Re: EURYSD D1 OBOS MA Osc with Break Even

Popov wrote:

This strategy can be improved if we trade it manually.

The main idea is that we don't need to close long position at bar close if we have to open long again at that time.

It's better to check the signal first and if the new trade has to be in the same direction, the exit order has to be canceled. In that way we can save one spread for these days.


What I've got all the fuses. Put file there historical bars for D1 where you tested please.

  • Clew
  • New member
  • Offline
  • Registered: 2011-07-02
  • Posts: 7

Re: EURYSD D1 OBOS MA Osc with Break Even

Hi,

The breakeven function doesn't seem to work correctly. For this strategy, i think it is more appropriate if FSB marks every bar as ambiguous.

It uses the "shortest" interpolation method all the time.
As quoted from here - http://forexsb.com/wiki/fsb/manual/interpolation_methods
"the price never goes back to a previous value. "  <- false

Try putting 1 pip breakeven, and use the "nearest" interpolation method. If at bar close the position is profitable, it will close there instead even if it is further away.
http://img98.imageshack.us/img98/5756/breakeveno.jpg

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Re: EURYSD D1 OBOS MA Osc with Break Even

FSB interpolates the bar according to the available info. You can improve the backtest correctnes by using intrabar scanning with shorter period or even tick backtesting. You can also apply Method Comparator to se results of all methods.

FSB sets Break Even exit order when it is in some calculation point. In the example above FSB will set BE when it is at bar low or bar closing. At that time, the BE order will be far alway and program will execute close at the bar close. Intrabar data and scaning will help in this case.

Popov's Website

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Re: EURYSD D1 OBOS MA Osc with Break Even

I see the problem now. I'll try to add calculation point at the Break Even activation price. It will force FSB to make decision where to go - to Break Even or to Bar Close.

Popov's Website

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Re: EURYSD D1 OBOS MA Osc with Break Even

The breakeven function doesn't seem to work correctly. For this strategy, i think it is more appropriate if FSB marks every bar as ambiguous.

Clew you were absolutely right.
It's the result of the backtest of the upper strategy with improved Break Even backtest:

http://s3.postimage.org/2o9gtsojo/11_08_2011_10_14_27.jpg

The improved FSB version will be uploaded soon.

Popov's Website

  • Clew
  • New member
  • Offline
  • Registered: 2011-07-02
  • Posts: 7

Re: EURYSD D1 OBOS MA Osc with Break Even

Hi again,

Thanks for the fix. smile

However i noticed something strange again.

For easier understanding, i changed the opening point to bar opening, and used 1 pip breakeven.
Previous bar value ticked for OBOS MA Osc. Optimistic interpolation method used.

Notice that profitable short entries will close at bar close and is marked as ambiguous, but profitable long entries will close at breakeven and is not marked as ambiguous.

The left half are filled with ! marks. None on the right half.
http://i1212.photobucket.com/albums/cc457/Clew_R/breakeven.jpg


Top Image with Breakeven. Bottom is without Breakeven.
http://i1212.photobucket.com/albums/cc457/Clew_R/breakeven4.jpg

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Re: EURYSD D1 OBOS MA Osc with Break Even

Clew, thank you for the priceless feedback.
It's true it looks suspicious. I'll check it now.
+1 from me

Popov's Website

  • Clew
  • New member
  • Offline
  • Registered: 2011-07-02
  • Posts: 7

Re: EURYSD D1 OBOS MA Osc with Break Even

The opposite happens for the Pessimistic Method. Just that position is closed at breakeven for both sides when profitable(correct). But Ambiguous bars only on the right half.  Shortest IM looks fine to me. 0 spread, 0 swap used.

http://i1212.photobucket.com/albums/cc457/Clew_R/breakeven5.jpg

And on the first position of every chart when using all the IMs, breakeven doesn't want to be activated.

Top Image = 1 hour chart. 4 pip difference between entry and bar high.
Bottom Image = 1 min chart. 1.1 pip difference between entry and bar high.
1 pip breakeven not activated.

http://i1212.photobucket.com/albums/cc457/Clew_R/breakeven6.jpg

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Re: EURYSD D1 OBOS MA Osc with Break Even

And on the first position of every chart when using all the IMs, breakeven doesn't want to be activated.

I noticed and fixed this already. Thank you!

Popov's Website

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Re: EURYSD D1 OBOS MA Osc with Break Even

The bug is fixed in Forex Strategy Builder v2.63.2 Beta.

It appears that this issue was old enough and is not a result of recently added Break Even code. But exactly Break Even made it obvious.

The fix may reflect on other strategies with ambiguous bars but it heavily changes the Break Even tests.

It seams that current methods of interpolation cannot properly calculate Break Even.

Both pessimistic and optimistic methods hit Break Even if it can be activated in the same bar.

This image shows the interpolation of one bar using both methods.

http://s3.postimage.org/afmr6ttw/Break_Even_interpolation.jpg

Popov's Website

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 7,988

Re: EURYSD D1 OBOS MA Osc with Break Even

Break even works well with backtesting with tick data.

http://s3.postimage.org/awobkl1g/Tick_Testing_Break_Even.jpg

Popov's Website