<- Back to Tasks

Mt5 Engine m1 vs every tick. Can't get m1 sqx tests to match m1 tests on mt5 terminal

Actually, "every tick real ticks" is matching pretty well with some strategies to MT5 terminal in every tick mode real ticks but I'd like to do some of my testing in m1 mode to save CPU. But, when I compare m1tests from SQX to m1 tests in MT5 it's not even close....

So in these screenshots, you can see the every tick real tick testing matches up OK for this limit order strategy running on USTECH100 dukas data....

 [see next comment...]









Attachments
Screenshot_240.png
Screenshot_240.png
(374.88 KiB)
Screenshot_246.png
Screenshot_246.png
(30.82 KiB)
Strategy 111071.sqx
Strategy 111071.sqx
(107.01 KiB)
Screenshot_241.png
Screenshot_241.png
(373.46 KiB)
Screenshot_247.png
Screenshot_247.png
(43.68 KiB)
  • Votes +2
  • Project StrategyQuant X
  • Type Bug
  • Status Fixed
  • Priority Normal

History

b
#1

bentra

10.10.2022 03:36

Task created

b
#2

bentra

10.10.2022 03:43

Attachment Screenshot_249.png added

Attachment Screenshot_248.png added

Screenshot_249.png
Screenshot_249.png
(331.55 KiB)
Screenshot_248.png
Screenshot_248.png
(485.70 KiB)
However here is m1 data tick simulation from SQX, from the tic data of the symbol.
b
#3

bentra

10.10.2022 03:49

Attachment Screenshot_251.png added

Attachment Screenshot_250.png added

Attachment m1-config-m1.cfx added

Screenshot_251.png
Screenshot_251.png
(369.22 KiB)
m1-config-m1.cfx
m1-config-m1.cfx
(46.05 KiB)
Screenshot_250.png
Screenshot_250.png
(323.91 KiB)
And here is m1 data tick simulation from SQX, from the m1 data version of the same symbol.
b
#4

bentra

10.10.2022 03:54

Attachment Screenshot_253.png added

Attachment Screenshot_252.png added

Screenshot_253.png
Screenshot_253.png
(507.66 KiB)
Screenshot_252.png
Screenshot_252.png
(349.40 KiB)
And here is m1 data tick simulation from SQX, from the m1 data version but with realistic gaps turned off.....
b
#5

bentra

10.10.2022 03:57

Attachment Screenshot_255.png added

Attachment Screenshot_254.png added

Attachment m1-config-tic-gapsoff.cfx added

m1-config-tic-gapsoff.cfx
m1-config-tic-gapsoff.cfx
(46.06 KiB)
Screenshot_254.png
Screenshot_254.png
(617.22 KiB)
Screenshot_255.png
Screenshot_255.png
(376.27 KiB)
And here is m1 data tick sim from SQX, from the tic data version of the symbol with realistic gaps turned off....


b
#6

bentra

10.10.2022 04:04

Attachment Screenshot_259.png added

Attachment Screenshot_258.png added

Attachment Screenshot_257.png added

Screenshot_257.png
Screenshot_257.png
(102.34 KiB)
Screenshot_258.png
Screenshot_258.png
(38.91 KiB)
Screenshot_259.png
Screenshot_259.png
(35.33 KiB)
And then in MT5 here is a test from the same dukas data in "ohlc m1" mode
b
#7

bentra

10.10.2022 04:17

Attachment Screenshot_262.png added

Attachment Screenshot_261.png added

Attachment Screenshot_260.png added

Screenshot_260.png
Screenshot_260.png
(34.59 KiB)
Screenshot_262.png
Screenshot_262.png
(16.19 KiB)
Screenshot_261.png
Screenshot_261.png
(30.75 KiB)
And here is MT5 test in every tick mode (but not real ticks)

b
#8

bentra

10.10.2022 04:21
So my point is it doesn't look like any of the m1 options match up very well between SQX and MT5 terminal. Surely some of it is because m1 ohlc mode in MT5 has variable spread and SQX doesn't support that for m1 mode but still we should be able to get closer than that.

Also, some clarity on which SQX m1 option is supposed to match m1 ohlc mode of MT5 would be nice too.
b
#9

bentra

10.10.2022 04:22
Voted for this task.
b
#10

bentra

10.10.2022 04:33

Attachment Screenshot_264.png added

Attachment Screenshot_263.png added

Attachment Retest strategies 6 - m1-tic-gapsoff.cfx added

Screenshot_264.png
Screenshot_264.png
(450.24 KiB)
Screenshot_263.png
Screenshot_263.png
(319.45 KiB)
Retest strategies 6 - m1-tic-gapsoff.cfx
Retest strategies 6 - m1-tic-gapsoff.cfx
(46.06 KiB)
Oops my dates got messed up in SQX.

Let's try again.

Here is m1 data tic simulation on the tic version of the symbol with realistic gaps turned off...
b
#11

bentra

10.10.2022 04:37

Attachment Retest strategies 6 - m1-tic-gapson.cfx added

Attachment Screenshot_266.png added

Attachment Screenshot_265.png added

Retest strategies 6 - m1-tic-gapson.cfx
Retest strategies 6 - m1-tic-gapson.cfx
(46.06 KiB)
Screenshot_265.png
Screenshot_265.png
(500.08 KiB)
Screenshot_266.png
Screenshot_266.png
(339.63 KiB)
Here is m1 data tic simulation on the tic version of the symbol with realistic gaps turned on...


b
#12

bentra

10.10.2022 04:46
So my point is it doesn't look like any of the 4 different m1 options match up very well between SQX and MT5 terminal. Surely some of it is because m1 ohlc mode in MT5 has variable spread and SQX doesn't support that for m1 mode but still we should be able to get closer than that, we are missing a lot of trades in all of the m1 tic sim option in SQX.

Also, some clarity on which of these 4 SQX m1 options is supposed to match m1 ohlc mode of MT5 would be nice too.
-m1 data tic sim on tic version of symbol with real gaps on

-m1 data tic sim on m1 version of symbol with  real gaps on

-m1 data tic sim on tic version of symbol with real gaps off

-m1 data tic sim on m1 version of symbol with real gaps off

E
#13

Emmanuel

10.10.2022 13:56
Voted for this task.
b
#14

bentra

18.11.2022 14:41
After more testing, these differences occur because of low data quality. When the quality is high, there is no difference. When using low-quality data, SQX appears to be handling the situation in a more realistic manner while MT5 handles it rather pessimistically. I have second thoughts about whether or not I even want it changed.
TT
#15

Tamas

09.01.2023 10:57

Status changed from New to Fixed

Thanks for the info about data quality . I am closing the task
Votes: +2

Drop files to upload

or

choose files

Max size: 5MB

Not allowed: exe, msi, application, reg, php, js, htaccess, htpasswd, gitignore

...
Wait please