Dear customer support,
I've been
writing for this bug for months on email and also to Tomas on the website. I've
been promised that it would be investigated and fixed in the 120 build and that
you know about it.
Actually I
can not understand why there is such a lack of support for the customers. We
have paid some money for the product and still we are on our own with it. With
all it bugs and problems. I've spent literally a whole year building and
rebuilding strategies each time after release.
It is a
great potential for the software and maybe Tomas does not have enough time to
help to all of us, but it does not discharge the fact that the customer support
level is extremely low for such a good product.
I am more
than happy to help with bug searching but i wish to receive an adequate
processing of the bugs or if it is not a bug converting it to the feature
(according to the user experience case e.a https://roadmap.strategyquant.com/tasks/sq4_4513 ).However it is not my task and duty to actively participate in bug
research.
Another example: Last QA (4.3.00) can not recognize instruments of SQX strategies (120). How could it be possible if the last update of QA was majory about SQX support? Or is it problem on my side?
Now I am
feeling useless trying so hard to find out the ways to fix unknown issues,
losing my money trying hard on software which is not reliable.
I deeply
regret that I use this source of communication, but I did not find how to
escalate this matter further.
I hope that
developing and management team will make some lessons from my critical feedback
and will make some changes in customer supports and debug process.
To the
matter of issue
Metatrader
5 netted engine always produces zero duration trades, so I can not build netted
strategies to use it on exchanges which are not allowed hedged EA’s. (build
120, screenshots attached)
I hope that
you resolve this issue soon.
Also I
truly hope that you treat my feedback in a constructive way as I want to make
the service and the product better.
Best wishes,
Valeriy
Status changed from New to Fixed
I'm very sorry to read your dissatisfied post. We try hard to investigate all reports added by our customers, but as our software is very extensive and there are many reports, it is very hard to manage all the requirements in time. This is a very specific branch where programming skills but also the ability to make tests and investigate strategy behaviour in different trading platforms are needed.
I personally investigated this problem and in build 121 the netting engine will be working better. I made some fixes around exit methods such as exit after bars, so it will be working properly in this upcoming release.
I builded some strategies with different settings and tested its results both in SQ and MT5 and I got 100% matches. The way indicators and other parts of strategy logic works is the same as in hedging mode, so there should be no problems with this.
Straight to the reported issue - The thing is that netting mode is a special kind of trading environment where trades are managed differently than in hedging mode.
Because of these differences, the trades made by a strategy must be interpreted in a little different way. I've spent a lot of time working on this kind of engine and comparing results with those in MT5.
I came to a conclusion that to be able to compare the trades between SQ and MT, I need to see every market entrance made by a strategy. For simple strategies that do not allow opening new orders when actively in a market position, it is simple and the tradelist could be made the same way as in hedging mode. But the problem is when the strategy for example sells some lots, then multiple times sells some more, then buys some etc.
To be able to check this kind of behaviour correctly, there must be all of the market entries listed. And that's when the zero duration trades show up.
It is just to know when the strategy made some trade and for what price and volume. When SQ is checking the builded strategies, it seems weird to it and it marks the strategies as faulty because of this.
We are thinking about ignoring this kind of check when using netting mode and we will definitely make some efforts to fix this issue in next releases.
For now I advise you to turn off the "zero duration trades" automatic filter, because otherwise none of the generated strategies will be saved.
Thank you for your report. We will do our best to improve our support.
Best regards,
Tomas
Attachment Strategy 33188.sqx added
Attachment working build - Copy.cfx added
If i switch off "too many trades closing at the same bar" I will receive curve fitted 'something' which I am afraid even test on MT 5.
https://www.screencast.com/t/5VPq8KF7F ; https://www.screencast.com/t/WXduhfxT6R ;
Added strategy and config file
I tried your attached config and it seems to be fine. But the attached strategy is weird. It does only short orders with big SL and very little PT level. If you retest it using higher precision, it will never be profitable.
But those types of strategies can be very easily filtered out using crosschecks. For example the one attached would not pass "Retest with higher precision" cross check.
We will fix the automatic checks and do some additional tests with netting engine until next release. Do you need some further assistance?
Attachment Strategy 11130.sqx added
Attachment Strategy 25176.sqx added
I've built for 2 hours in netted mode on 30m and have only 6 strategies. https://www.screencast.com/t/54RRel46bY https://www.screencast.com/t/5qOb0oqfw ;
I've turned off "too many trades closing at the same bar' and 'zero duration trades'
I've also turned on "retest with higher precision'
Two of strategies looks weird in equity curve. Attached here https://www.screencast.com/t/nt00tX1uns
Still I can not say that I am ready to use it on netted exchange and that netted mode is solid.
Can you help me and say why it is so hard to SQX to build a new strategies in netted mode?
Build121
I loaded your attached strategies into Retester and found out there is a bug in MAE/MFE computation, so thank you, I haven't noticed this problem so far.
It's fixed and it will be working well in the next build. It was just a matter of bad dividing. After the fix, the curves look normal