Sampling H4, D1 Irreguarities result in MT4 data mismatch
Posted: Tue Jul 16, 2019 9:51 pm
See attached XLSX for detailed analysis of an MT4 data mismatch error isolated to Tickstory H4 Sampling
a) download TS data
b) export TS data to MT4
c) using MT4 history centre export M1, H4, D1 to CSV
d) assemble all the data in attached
Sheets
Log - shows the MT4 error and the TS download parameters
Comp - shows the differences in H4 high prices and D1 closing prices when manually creating same with M1 data
M1.M5,M15,M30,H1,H4,D1 - show the MT4 history centre export CSV content (zeroed in on just that area around Jan 25, 2008
Data Used: GBPJPY 2008.01.25
Bar in Question : 20:00
Auxiliary analysis shows that this tends to happen on weekends only, that is on Fridays before the weekend. This suggests some sort of glitch on TS algo for H4 and D1 as recreating TS sampled M5,M15,M30,H1 from M1 TS checks out perfectly.
Also confirmed that when i take that TS M1 hst and use the MT4 PeriodConverter Script, it samples correctly resulting in NO data mismatch errors.
Please advise.
2019.07.17 Update
-Further reading suggests Dukas Copy's MT4 data is GMT+3,+2 timestamped though Tickstory manual says the tickdata is GMT +0 timestamped.
-Running the Tickstory Export WITH GMT=0 (no offset) gives NOT mt4 data mismatches.
-The reason for exporting to GMT+2/3 (DST on/off) is that my broker uses that timezone setting for the endof the daily bar aka NY 5pm close.
So how can i eliminate the data mismatches ? (some flag in Tickstory export and/or settings or discard all but M1 from tickstory and run period converter ?) Without knowing what exactly is going on it's difficult to trust even the M1 data ...
a) download TS data
b) export TS data to MT4
c) using MT4 history centre export M1, H4, D1 to CSV
d) assemble all the data in attached
Sheets
Log - shows the MT4 error and the TS download parameters
Comp - shows the differences in H4 high prices and D1 closing prices when manually creating same with M1 data
M1.M5,M15,M30,H1,H4,D1 - show the MT4 history centre export CSV content (zeroed in on just that area around Jan 25, 2008
Data Used: GBPJPY 2008.01.25
Bar in Question : 20:00
Auxiliary analysis shows that this tends to happen on weekends only, that is on Fridays before the weekend. This suggests some sort of glitch on TS algo for H4 and D1 as recreating TS sampled M5,M15,M30,H1 from M1 TS checks out perfectly.
Also confirmed that when i take that TS M1 hst and use the MT4 PeriodConverter Script, it samples correctly resulting in NO data mismatch errors.
Please advise.
2019.07.17 Update
-Further reading suggests Dukas Copy's MT4 data is GMT+3,+2 timestamped though Tickstory manual says the tickdata is GMT +0 timestamped.
-Running the Tickstory Export WITH GMT=0 (no offset) gives NOT mt4 data mismatches.
-The reason for exporting to GMT+2/3 (DST on/off) is that my broker uses that timezone setting for the endof the daily bar aka NY 5pm close.
So how can i eliminate the data mismatches ? (some flag in Tickstory export and/or settings or discard all but M1 from tickstory and run period converter ?) Without knowing what exactly is going on it's difficult to trust even the M1 data ...