• Registered: 2017-12-01
  • Posts: 19

Topic: Bug or is it just me?

Hi, I usually make manual strategies using FSB. I am trying to use "human" logic within FSB limitations. The thing is that using human logic, I often find strategies with almost perfect curves but the problem is that when backtesting them in MT4, they almost all fail brutally. Once I have posted I have found a bug using Seady bands as closing logic. I have to say that many "band" indicators in FSB might be buggy.

This time I have found a simple strategy that shows good curve but again in backtest fails badly.
I don't know is it just me or I have bad settings in FSB?
Let me say that I use 1000€ initial balance, leverage 1/100, micro account, spread 25, and the rest of the settings like in MT4 demo account. I use SQ Tick downloader that downloads Dukascopy data that I import in FSB.

Here is my latest strategy. Please make a comment.

https://s17.postimg.org/o01jn5smj/Time_grid.jpg

Market
     Symbol  -  EURUSD
     Period  -  M1

Trading Size
     Maximum lots  -  2
     Trading units  -  lots
     Entry units  -  2
     Reducing units  -  1

Strategy Properties
     Same direction signal  -  Nothing
     Opposite direction signal  -  Reduce
     Stop Loss  -  None
     Take Profit  -  None
     Break Even  -  None

[Opening Point of the Position]
     Bar Opening
     Enter the market at the beginning of the bar
     Base price  -  Open


[Opening Logic Condition]
     Time Grid
     Bar opens at round 15 min
     From hour (incl.)  -  0
     Until hour (excl.)  -  24

     Logical group  -  A

[Opening Logic Condition]
     Time Price Channel
     The position opens above Upper Band
     Base price  -  Open
     Hour  -  7
     Minutes  -  0
     Price deviation [point]  -  5
     Use previous bar value  -  No

     Logical group  -  A
     Signal shift  -  0
     Signal repeat  -  0
     Indicator symbol  -  Default
     Indicator period  -  Default

[Closing Point of the Position]
     Bar Closing
     Exit the market at the end of the bar
     Base price  -  Close


[Closing Logic Condition]
     Time Price Channel
     The bar closes below Upper Band
     Base price  -  Open
     Hour  -  7
     Minutes  -  0
     Price deviation [point]  -  5
     Use previous bar value  -  No

     Logical group  -  a
     Signal shift  -  0
     Signal repeat  -  0
     Indicator symbol  -  Default
     Indicator period  -  M15

Thanks

  • Registered: 2017-01-14
  • Posts: 80

Re: Bug or is it just me?

hello arsnuova, try to use the data of your broker and, in the absence of out of sample, cut the data manually using "beginnig / updating"; created the strategy uses the data in their entirety to eliminate overfitting. Finally try with the broker

  • Registered: 2017-12-01
  • Posts: 19

Re: Bug or is it just me?

D5QM54S7 wrote:

hello arsnuova, try to use the data of your broker and, in the absence of out of sample, cut the data manually using "beginnig / updating"; created the strategy uses the data in their entirety to eliminate overfitting. Finally try with the broker

Thank you. I do it the way you described. I never use OOS but manually choose a beginning.
I would use my brokers data (xm) but it doesnt't have enough data so I use Dukascopy as find it most reliable.
I am  not a trader, I am only interested in forex theory. As far as I know open and close prices are allways the same, right? Usin a strategy based on bar opening and bar closing, doesn't it mean that different brokers data don't make any difference?

The strategy I posted shows the positive equity line no matter which date it starts. Pleae try it in FSB and let me know. Thanx

4 Reply by D5QM54S7 (edited by D5QM54S7 2018-01-20 19:13:28)

  • Registered: 2017-01-14
  • Posts: 80

Re: Bug or is it just me?

I have more accounts including the one you mentioned, it is not recommended to use data from other brokers, in my case the experience was very bad. I often do not even try the ea on mt4 but I load it directly in real. I found differences not only between different brokers but also between demo and real of the same broker. Little timeframe = big sensitivity -> if you want to study this timeframe you have to use perfect data

  • Popov
  • Popov
  • Lead Developer
  • Offline
  • From: Bulgaria
  • Registered: 2006-11-30
  • Posts: 8,024

Re: Bug or is it just me?

MT cannot backtest (or even trade) correctly at bar closing because it lack of functionality of detecting that point. Theoretically, a Bar Closing must happen at the exact time when the bar period expires. However, MetaTrader cannot make that simple math. If you monitor a chart closely, you will notice that MT starts on a new tick after the bar open time. Even worse, on very slow markets MT completely misses bars if it happens that there is no ticks on that period. On unknown reason MT doesn't recognizes bar closing correctly on a completely controlled environment in their historical tester. I personally reported that bug to them 2-3 years ago. EA Studio uses only Bar Open in order to provide the best user experience for our clients.

Please examine your strategy work in MT and you may find the place where the discrepancy happen. It is very unlikely to have a bug in the FSB Pro or in EA Studio backtesters. It is possible to have problems in a specific custom indicator but the included "band" indicators are available in FSB from 10 years and we may assume they work well. 

If you think that there is a bug in a specific indicator, please make really simple strategy that utilizes only that indicator. You will easily find if it works well.

Popov's Website

  • Blaiserboy
  • Blaiserboy
  • Junior Part-Time Aspiring Space Cadet
  • Offline
  • From: Toronto
  • Registered: 2010-09-06
  • Posts: 2,849

Re: Bug or is it just me?

I think the strategy that you posted is one minute.
I suggest that for testing purposes you move up to 60 minute as there are many things that can happen with one minute such as your internet connection, your computer speed, latency, the broker's trading work load.....
You may have several things happening at once on your computer to cause delay in execution.

One minute is pretty fast and for you to get good results I strongly suggest that you up the time frame and then you can know more for sure.

Blaiserboy's Website