SQX v SQ3.8.2 – Strategy Creation Count Performance

Hi Mark

I have started to use SQX in the last week, since its commercial launch. My first task was to setup a SQX setting template as similar as possible to one I have previously used with SQ3.8.2. Then run SQX on one PC and SQ3.8.2 on another (very similar) PC for a similar period of time, to compare strategy creation count performance. My setup was as follows:

• SQ3.8.2 run on PC; Windows 10 Pro 64bit; Intel i7-4790 4-core @ 3.60GHz CPU; 32Gb RAM
  • SQ3.8.2 setup as 5x separate instances: 1-thread per instance
  • Number of strategies created in ~24hr period: 132 (across each instance: 16; 20; 52; 12; 32)
• SQX run on PC; Windows 10 Pro 64bit; Intel i7-6700K 4-core @ 4.00GHz CPU; 32Gb RAM
  • SQX setup as a single instance, with performance set to: Max Performance; 8-thread
  • Number of strategies created in ~24hr period: 4

As you can see SQX was given a head-start over SQ3.8.2: slightly higher performance PC; 8-threads v 5-threads. Yet strategy creation count performance was far better with SQ3.8.2, than SQX.

Data symbol used: NZDUSD; M1; UTC+2DST; 2003.8.4 to 2018.10.19; timeframe H4

I realise that it is difficult to exactly match the same settings for both platforms, but I tried my best, including:

• Use SQ3 architecture setting in SQX (to act similar to SQ3.8.2)
• Use only 1x Island in Genetic Evolution (to act similar to SQ3.8.2)
• All Automatic Filters deselected (to act similar to SQ3.8.2)
• With SQ3.8.2 I used a filter in Genetic Evolution (dismiss Net Profit [IS & OOS] <$0), which I initially also enable in SQX (accept Net Profit [IS & OOS] >$0), but subsequently disabled after 12hrs, as SQX had generated Zero strategies

The only significant setup difference I am aware of, was:

• SQ3.8.2 data Test Precision set to: 1 minute data (my usual setting)
• SQX data Test Precision set to: 1 minute data tick simulation (‘1 minute data’ setting currently not available in SQX)

Please see attached copy of both my SQX and SQ3.8.2 setting files. (Is this the 'Config' file you request? If not please confirm location and name of 'Config' file that you need)

Please can you look at these settings files and comment as to what may be the cause of the significant difference in performance I am currently seeing between SQX and SQ3.8.2? Is this the result of a bug in SQX, or an issue with my SQX settings?

Thanks

AC1962