new unmatched data errors from existing history files

 

Hi all,

I do understand that much has been written on this topic, but I'm struggling to line up my exact problem with the rest of the problems and solutions covered already.

I'm testing some EAs in MT4 using a set of .csv history files purchased online in June 2020 from a reputable source.

I'm testing all FTSE timeframes. I've been successfully running backtests with these files hundreds of times on all timeframes, and with no mismatched errors until this week.

I have my old backtest reports, showing the millions of bars processed with zero errors, on all timeframes from 1 minute to daily.

This week, I have numerous errors of mismatched prices and volumes. First I re-imported all the history numerous times, then I started checking the files carefully. The modified dates on the files are still 4th June 2020 and I can go back to the original .zip file and have checked the integrity there too, so the data I am using seems to be what I originally used for all previous tests.

From looking at individual errors forensically, I can see that indeed the HIGH value on the daily chart is exceeded by some bars on the 1 minute file on the same day.

From the daily file, showing the high of 5795.5 here

04/05/2020;00:00:00;5666;5795.5;5645;5733.5;121191


And examples from the 1 minute file, showing the values exceeding 5795.5 on the same day:

04/05/2020;23:43:00;5805.5;5809;5805.5;5809;5

04/05/2020;23:46:00;5804.5;5808.5;5804.5;5808.5;10

04/05/2020;23:54:00;5808.5;5818;5808.5;5818;5

04/05/2020;23:55:00;5814;5814;5813.5;5813.5;7

04/05/2020;23:56:00;5816.5;5816.5;5816.5;5816.5;1


So - it looks like the errors are indeed errors and should always have been reported. All this information seems to suggest these two things: 1) that I was sent bad data by this provided AND 2) that I've been running tests without any of the errors being reported until now. I cannot believe that both of these things are true!

Does anyone have further background on this problem and possible solutions?

Thanks very much. 

 
Ask your “reputable source” how to create all timeframes from their CSV file(s).
Reason: