Trading isn't consistent with the signals

[jollydragon]

Please let me update this post with the latest status.
I've updated the code files also as the latest versions as enclosed.

According to the EA, an old order should be closed and a new order should be open just 1 bar later than the signal.
However, the back testing still isn't consistent with the indicator signals and please refer to the screenshot below.
This issue only happens seldom during back testing, about 10% of the indicator signals. However, it exists and not reasonable.

trade joy.mq4

(11.04 KiB) Downloaded 659 times

Solar Joy.mq4

(6.65 KiB) Downloaded 637 times

[tickstory]

Hi Jollydragon,

Unfortunately we don't provide support for trouble-shooting EAs so hopefully someone else from the Tickstory community may be able to assist. Just bear in mind that Tickstory is exporting data - the subsequent trading logic is entirely dependent on your EA code. If you believe there is a specific issue with the data that Tickstory is exporting we need more specific information for us to investigate further.

Thanks.

[jollydragon]

Hi Jollydragon,

Unfortunately we don't provide support for trouble-shooting EAs so hopefully someone else from the Tickstory community may be able to assist. Just bear in mind that Tickstory is exporting data - the subsequent trading logic is entirely dependent on your EA code. If you believe there is a specific issue with the data that Tickstory is exporting we need more specific information for us to investigate further.

Thanks.

Dear Tickstory,

After almost one month of study, I can confirm it's the issue of TickStory now.
I also updated my previous post above with latest status and code files.
My back test is completely consistent with my indicator if not using TickStory, but the data of broker own.
You can try with the files for the testing of EurUsd M30. Maybe you'll find the issue within only 10 trades.
Please let me know if you need any additional information.

By the way, I also found there is only 1 tick per bar for periods of M1, M5, M15 on EurUsd chart with TickStory back test. How can the report finally achieve 99.9% quality?

[tickstory]

Hi Jollydragon,

As mentioned, we simply don't have the capacity to debug through each of the EA code samples we receive. Not only is it extremely time-consuming, but while custom code is still in the picture there are too many variables that make it hard to identify where the issue lies.
If you think there is a problem with the data, could you please isolate the details so we are able to work with actual exported data rather than an EA. There can only be two causes if Tickstory is giving you the wrong back-test results: 1) The exported data is incorrect or 2) the MT4 broker information is incorrect. In both cases, you should be able to identify the point at which your EA is not taking a correct trade. From there, it should be possible to pinpoint a specific date/time of the bar data and/or parameters that are incorrect which we then can investigate further.

Thanks.

[jollydragon]

Hi Jollydragon,

As mentioned, we simply don't have the capacity to debug through each of the EA code samples we receive. Not only is it extremely time-consuming, but while custom code is still in the picture there are too many variables that make it hard to identify where the issue lies.
If you think there is a problem with the data, could you please isolate the details so we are able to work with actual exported data rather than an EA. There can only be two causes if Tickstory is giving you the wrong back-test results: 1) The exported data is incorrect or 2) the MT4 broker information is incorrect. In both cases, you should be able to identify the point at which your EA is not taking a correct trade. From there, it should be possible to pinpoint a specific date/time of the bar data and/or parameters that are incorrect which we then can investigate further.

Thanks.

After studying further, I did find something. But need your professional support to dig out the root cause.
Please refer to the screenshot below that the non-consistency issue doesn't happen if using "visual mode".
By the way, you can see there is 3 hour difference between TickStory and MT4 for a same bar.

P.S. Below are the screenshots of non-consistency issue if without ticking "visual mode". Just for reference.

[tickstory]

Hi,

Attached is a screenshot of the visual tester for the date you mention - 17th June 2013 - for EURUSD in M30 time-frame and using 'MACD Sample' EA. No time shift was applied to the data export so all times are in UTC (I note you haven't mentioned whether you have applied any time-shift, so I'll assume this must be the reason for your 3 hour difference). As you can see, the 8am bar matches exactly with the raw data.
The chart has been run through the back-tester in visual mode and for the bar you mention, it corresponds to the underlying MT4 history center for M30 (see screenshot).

It is still not clear how we can assist you further without debugging your indicator. As such, you will need to provide sufficient information for us to reproduce the issue without requiring us to debug custom code. The first step is to isolate all custom code that may be causing the issue you are having and, if you are suggesting that Tickstory data is incorrect, identify which part of that data is incorrect (via print statements, etc). This is what we would need to do if we were to assist further (and again, as mentioned, we do not have the capacity to assist in this way). If you can clearly demonstrate there is an issue with the exported data then we can investigate further.

Thanks for your cooperation.

[jollydragon]

Hi,

The chart has been run through the back-tester in visual mode and for the bar you mention, it corresponds to the underlying MT4 history center for M30 (see screenshot).

It is still not clear how we can assist you further without debugging your indicator. As such, you will need to provide sufficient information for us to reproduce the issue without requiring us to debug custom code.

Thanks for your cooperation.

Hi, please read my previous post again that no issue in visual mode, the issue happens in normal testing mode.
With my EA and indicator enclosed in the first post, you can reproduce the issue completely. Let me remind with the issue situation again:
. No issue-use tickstory in visual mode;
. No issue-not use tickstory, but MT4 alone;
. with issue-use tickstory in normal testing mode(not in visual mode)

[tickstory]

Hi,

Please let us know when you have reproduced the issue without requiring us to debug custom code. If there are differences between visual and non-visual mode, then a quick print statement of the raw values (i.e. the OHLC data points - not your custom indicator) should be able to demonstrate this.

Thanks.