SQ Xのアップデート

ここからSQ Xのアップデートやニュースについて話し合うことができます。.

このトピックは「SQ4 Early Preview」から名称変更されました。古い、もはや関連のない投稿は削除され、アーカイブされます。.

ポイントは、古い、関連性のない内容からフォーラムを整理することです。.

stopping at checking data – that’s strange, we installed it on multiple computers during tests and it always worked without problems.

Have you made clean installation or updated an existing one?

Can you please create new bug task and attach your log? It is located in /user/log/StrategyQuant

Or attach the log here.

Anyone else is experiencing this error?

I updated from 106 without problems

I tried to reinstall after no problem.This version is very good performance~~

(Attachment is previously updated from RC5, execution error message)

Thanks a lot~

build 107 is not better, again the strategy builder’s low cpu usage is not fixed since RC5,

the wizard is a surprise ,  why the UI was totally changed when we were ok with the old one , the interface is not a issue , the issue is the large change on that cause you to fix lots of low low lever bugs, but the feather of run the test is a big plus .

i also try to ran the old wizard attached but dashboard bottoms  not working , load, save is not working , no response.

anyhow thank you Mark for your work.

yourrodstaffさんはこう書いた：

build 107 is not better, again the strategy builder’s low cpu usage is not fixed since RC5, the wizard is a surprise , why the UI was totally changed when we were ok with the old one , the interface is not a issue , the issue is the large change on that cause you to fix lots of low low lever bugs, but the feather of run the test is a big plus . i also try to ran the old wizard attached but dashboard bottoms not working , load, save is not working , no response. anyhow thank you Mark for your work.

I agree with yourrodstaff, the GUI is worst, too much slower that the 106 build… The dark skin have a lot of glitches (not in the last 106 build) and for me also is going more slower than the 106 build (tested on a dedicate server with 8 xeon cpu, 80 core, 160 thread)

The implementation over linux is a great idea, is written in java so i don’t think it’s very difficult recompile all for linux, anyway i still don’t understand why use this heavy slower UI, i think a lot of user (me included) run this software over dedicated server, where the gpu is something integrated with very slow power…
Can you implement a parallel version with a light GUI, not managed from chromium or something like that? It would be wonderful. Thanks for all your time and work.

Me too. Stuck at checking data for hours.

(The \usr\log\StrategyQuant folder is empty)

I have a different kind of challenge. While previous builds were crashing and doing nothing at all, with a bit of guidance from Mark, I have been able to generate some strategies with Build 107. It is stable – very much so.

My problem is managing memory. It is a problem i have reported before, and still exists. My question here is how I clear the memory (or reset)  when it maxis out, like the attached screenshot.

Your advice is much appreciated.

yourrodstaffさんはこう書いた：

build 107 is not better, again the strategy builder’s low cpu usage is not fixed since RC5

low CPU usage? that’s strange, how do you have it configured? SQ X can use 100% of your CPU

RNG wrote:

I agree with yourrodstaff, the GUI is worst, too much slower that the 106 build… The dark skin have a lot of glitches (not in the last 106 build) and for me also is going more slower than the 106 build (tested on a dedicate server with 8 xeon cpu, 80 core, 160 thread)

we are working on dark skin, it woudl help if you could create tasks with screenshots for these glitches, if you haven’t done so.

I don’t understand what’s the issue with Chromium, especially on a powerful computer such as yours. Chromium is fast and it takes only minimum CPU even on slow laptop. It definitely doesn’t slow SQ down.

What exactly feels slow in SQ UI?

Also, in my tests with the same settings build 107 is faster than build 106.

GaryAitcheson wrote:

Me too. Stuck at checking data for hours. (The \usr\log\StrategyQuant folder is empty)

infinite Checking data is a bug, I’m sorry for it.
It is caused by some issue in update process, we are looking at it. Clean installation works fine.

Jojo wrote:

I have a different kind of challenge. While previous builds were crashing and doing nothing at all, with a bit of guidance from Mark, I have been able to generate some strategies with Build 107. It is stable – very much so. My problem is managing memory. It is a problem i have reported before, and still exists. My question here is how I clear the memory (or reset) when it maxis out, like the attached screenshot. Your advice is much appreciated.

SQ used 14 GB of RAM in your case, how big is your databank?

Usually most of the memory is used by strategies in databank. To clean it up just save the strategies from databank and clear it.

If your databank is empty and you still have big memory usage could you open a task for it and attach your setting there?

Mark or anyone who has noticed this :

i have been testing StrategyQuant X build 107 and most of the system generated by default settings are very much Curve Fitted . I have tried several building blocks and even entire blocks but still almost all systems generated are curve fitted . StrategyQuant 3.8.2 was not curve fitting as much this new version StrategyQuant X build 107.

Have you noticed the extreme Curve fitting with StrategyQuant X build 107

Mark Chromium is fast as you want but never fast like a native GUI, i think the entire comunity agree with me when i say that the old GUI of version 3.8.2 was more responsive and fast than this new one.

developeralgo222 wrote:

Mark or anyone who has noticed this : i have been testing StrategyQuant X build 107 and most of the system generated by default settings are very much Curve Fitted . I have tried several building blocks and even entire blocks but still almost all systems generated are curve fitted . StrategyQuant 3.8.2 was not curve fitting as much this new version StrategyQuant X build 107. Have you noticed the extreme Curve fitting with StrategyQuant X build 107

no, I haven’t noticed it, and I don’t think it is true. How do you recognize if system is curve fitted?

SQ X gives you many more tools to prevent curve fitting than SQ 3.

RNG wrote:

Mark Chromium is fast as you want but never fast like a native GUI, i think the entire comunity agree with me when i say that the old GUI of version 3.8.2 was more responsive and fast than this new one.

you are right that the UI based in browser feels different than native one, and could be less responsive, but it also allows us to do new things. We don’t plan to make native GUI.

What exactly are the main problems you have with the UI?

Generating the strategies is mostly a matter of setting it up and letting it run.

マーク・フリックはこう書いた：

infinite Checking data is a bug, I’m sorry for it. It is caused by some issue in update process, we are looking at it. Clean installation works fine.

If I do a new install, what folders do I copy over to keep my data, settings, etc. ? Is it the just the ‘User’ folder ?

developeralgo222 wrote:
Mark or anyone who has noticed this : i have been testing StrategyQuant X build 107 and most of the system generated by default settings are very much Curve Fitted . I have tried several building blocks and even entire blocks but still almost all systems generated are curve fitted . StrategyQuant 3.8.2 was not curve fitting as much this new version StrategyQuant X build 107. Have you noticed the extreme Curve fitting with StrategyQuant X build 107
 

no, I haven’t noticed it, and I don’t think it is true. How do you recognize if system is curve fitted?

SQ X gives you many more tools to prevent curve fitting than SQ 3.

———-

my experience is:

For example I generate 5000 Strategies with Duka-Data EURUSD H1 on SQ 4.X

After my filtering with  dukadata only 200-300 Strategies are left.

In the next Step I filtered with AsikuryData.

After that no strategies are left.

=> This can be curvefitting or a bug in the backtest engine.

I checked this with Multitimeframe H1/Daily and single Timeframe. The result ist nearly the same. No strategy left my tests.

I don´t trust the backtestengine (tick simmulation). Ok I know we need more tests to check this.

Possible we can test the engine:

Generate simple strategies. With SQ 3.8.2 and SQ4.X and generate metatrader backtests. I think its difficult to test

it and find bugs.

I filter with unseen data, Profitfaktor, Retdd, Stagnation, Symetrie