Corrupted history detected (MT5)

[zirtico]

I’ve downloaded and re-downloaded a bunch of data from 2007-2023 for several symbols. I keep getting the error for “corrupted data” in the MT5 Journal. The chart does look a little weird, with bars overlapping and looking unnatural. I deleted all the data and did a fresh download and import and it’s still the same problem. When I test on data after 2011, I don’t seem to get the error. I just bought the premium subscription so this is a little disheartening…

[tickstory]

Hi Zirtico,

Could you please confirm you have followed this guide to import your data? If you are still having issues after that, could you please provide a screenshot of the error you are getting?
And finally, could you please provide us with the email address you used when registering so we can provide you with the appropriate support according to your subscription.

Thanks.

[zirtico]

I can confirm that was the process I've used. I've used Tickstory data successfully before, so this is not my first time.

This is what the error looks like:
NM 2 19:37:15.823 History NZDUSD 2016.09.21 23:01:00: corrupted history detected (s:-73370, o:73433, h:+48, l:-123, c:-117 -- tv:63, rv:11250111)

I'm happy to share my email address, but can I share it privately rather than on a public forum?

[tickstory]

Hi Zirtico,

Re: your email address, just reply here and we will ensure it does not appear publicly. We will also update your profile on the forum so we can identify your license.

Regarding the error you are getting, it seems that MT5 is suggesting that the data has invalid pricing information (negative low and closing price etc?). We've checked our NZDUSD minute data for the date/time mentioned and cannot see any such negative prices appearing in the data (we've checked in both Tickstory's Data Inspector and the MT5 exported CSV). You might want to check your own raw CSV file data file to make sure this is also the case.
You might try and omit the 2016 data from your export and see if that helps.

Outside of this, there are various mentions of this error on MT5 forums however the suggestion is that there is something wrong with MT5's data repository which necessitates deletion and/or trying a different MT5 terminal. If that's the case, the Metatrader forums might be a better source of identifying the issue.

Hope this helps.

[zirtico]

Hi, yes, I don't think it was just for NZDUSD. I got different sets of numbers for multiple pairs. The bars on the chart seemed to overlap and the price action did not look as clean as I'm used to seeing for high liquidity pairs. If this was USDRUB or something like that, I'd think it was more in line with my expectations. I also deleted the CSVs and all the MT5 data and did a full redownload of all pairs' data and still got the same error.

[tickstory]

Hi Zirtico,

I'd highly recommend you attempt to export just a subset of the latest data from say the latest 12 months of the NZDUSD pair and see if this addresses the issue. While the exported data from Tickstory looks correct, I can only assume that there's some sort of internal issue with the MT5 database.

Hope this helps.

[zirtico]

I think it has to do with date rather than with pair. Have a look at the error below, attached. I think you're right, most of the data does look OK. I just did another full re-download. I'm backtesting an EA over a long period of time (multicurrency to maximise trade sample size), so I do need as much good data as I can get. A couple of bars of bad data shouldn't be a problem as long as the rest of the optimization works ok.

[tickstory]

Hi Zirtico,

Are you saying the error is because of the actual date output or related to the specific dates giving some problems?
Did you manage to just export 12-months and see if that gives any problems?

Thanks.

[zirtico]

I can export 12 months of data without a problem in many cases. In this case it seems to be specific to that date.

[tickstory]

Hi Zirtico,

We've tried the following:

1) Exporting NZDUSD 1-minute data from Jan 2016 to Jan 1017 into MT5 and then testing with "Every tick" method. There was no error.
2) Exporting additional tick data from the same period and then testing with "Every tick with real ticks". Again, there was no error.

Since we don't know under what circumstances this error can occur, we are limited in what we can tell you about it. It sounds more like an internal application problem rather than a data-point related issue as it would be logical to present any input data integrity errors during the import process.

This error really is something that Metaquotes should be able to advise you more on. If you do get a solution from them, it would be appreciated if you can share the cause on this thread so others will know.

Thanks.