The downloaded historical data does not match the chart historical data.

[HalTech]

For NinjaTrader Continuum data. The downloaded historical data does not match chart historical data.

In each download of data below, the NT minute folder in the NT db folder was deleted before NT was started. This was to ensure a clean set of data.

I started NT and closed all charts and windows.
Global merge policy = Merge back adjusted
I connected to NinjaTrader Continuum data feed.
I went to the Historical Data download manager and verified there was no minute data.
In the Historical Data download manager I selected MES 12-23
Download dates = 05/05/2019 to 09/28/2023
I downloaded the data.
I closed NT and then saved the Minute folder in the NT db folder to a separate storage area.

I deleted the Minute folder in the NT db folder (to delete all of the minute data).
I started NT with no charts or windows open.
Global merge policy = Merge back adjusted
I created a new 5 minute chart MES 12-23
I set the days to load to 2100 (this corresponds with 05/05/2019 above)
I connected to NinjaTrader Continuum data feed.
The chart data loaded and the minute folder in the NT db folder filled with data.
I closed NT and then saved the Minute folder in the NT db folder to a separate storage area.

After changing NT to Merge Non back adjusted I did the same downloads.

So there are 4 different sets of minute data.
Merge Back adjusted
1 - Downloaded with Historical Data manager
2 - Downloaded by opening chart with 2100 days of data

Merge Non Back adjusted
3 - Downloaded with Historical Data manager
4 - Downloaded by opening chart with 2100 days of data

I then used a file compare utility (binary compare) to compare the Downloaded data to the data downloaded with the chart.
Here are the comparisons:

1 to 2 Downloaded-back adjusted to Chart-back adjusted and the results are:
997 same files
528 difference files
12 files only in Downloaded data folder
20 files only in Chart data folder

1 to 3 - Downloaded-back adjusted to Downloaded-Non back adjusted and the results are:
998 same files
534 difference files
5 files only in Downloaded data folder
10 files only in Chart data folder

1 to 4 - Downloaded-back adjusted to Chart-Non back adjusted and the results are:
691 same files
830 difference files
16 files only in Downloaded data folder
24 files only in Chart data folder

2 to 3 - Chart-back adjusted to Downloaded-Non back adjusted and the results are:
1534 same files
0 difference files
11 files only in Downloaded data folder
8 files only in Chart data folder

2 to 4 - Chart-back adjusted to Chart-Non back adjusted and the results are:
701 same files
833 difference files
11 files only in Downloaded data folder
11 files only in Chart data folder

So it appears that the best match is 2 to 3.
The Non back adjusted data downloaded with the Historical Data download manager is most equivalent to the Chart downloaded back adjusted data.
So since NinjaTrader defaults to Merge back adjusted, then the best data to download to match the chart data is the Merge Non back adjusted data.

I have repeated the downloads multiple time to be sure that the results are the same.

So why is the data that the chart downloads, different from the data that is downloaded with the Historical Data manager?

[bltdavid]

I then used a file compare utility (binary compare) to compare the Downloaded data to the data downloaded with the chart.

So why is the data that the chart downloads, different from the data that is downloaded with the Historical Data manager?

There is no such thing as 'downloaded' vs 'chart' historical data,
it's just historical data.

AFAIK, the '.ncd' file format has not been made public.

Did you sync your PC clock's before each download, just to
make sure?

What file compare utility did you use?

Did you convert the '.ncd' files to text files and see if the
binary differences carried over to text differences?

CC > Tools > Export > Historical data
converts '.ncd' files to '.txt' files

[EDIT: 'CC' is short for 'Control Center']

You went to a lot of trouble to identify differences in the files,
but did you see any visual differences in the data on the charts?

Since each minute '.ncd' file corresponds to a specific day,
when you identified a file difference, did you go to that day in
the chart and see any differences in the charts for that day?

Did you try with a different instrument, such as 'ES'?

-=o=-

In any event, unless the bars on the charts are affected, I'm
not sure it's something worth really caring about. I mean, this
is such an esoteric experiment, why even bother to do all this?

Did you have some other problem that prompted you make
these file comparisons?

I'm just curious, what were your motivations?
Is there an underlying issue that you're trying to solve?
Or, is this more an academic pursuit?

I'm presuming the differences are benign, and perhaps related
to timestamps or padding ... that's why I'm curious about any
differences you identified in how the chart data was drawn,
and did any of binary differences carry over when you exported
the historical data to text files.

But, yeah, I would tend to agree ... it is interesting.

[HalTech]

Thank you bltdavid I appreciate your information. I will export the data to text files to see any differences.
I appreciate your time.