Historical Data - Mismatch between 1m and 3m candles

edited June 2021
Hi,
I am working with Historical Equity data using your API. Reading previous discussions, it seems you do create 3m candle from 3 (1m candle). Given That I was checking and data seem to mis-match. Can you please help be clarify on the reason or logic, or I am missing something. Example

For Date: 2021-04-013
Symbol: Reliance

As you can see..the volume are not matching. 9:15 and also sum of 9:16-9:18.
Even the Open, High, low and Close from 1m*3, are not correct in 3m candle.

This seem to be the case across different scrips and dates.
What I am missing here?

Also:
1. Can you please guide me which is the closest correct candles (1m or 3m) to use for back testing.
2. Is data good (close to real) for last 5 years ? or do you have any cut off date - from which data is better?
Ps: Adding data also for clear view

3m - historical candle data
time open high low close volume
13-04-2021 09:15 1924 1933.7 1924 1931.05 397741
13-04-2021 09:18 1931.05 1931.25 1928.35 1929.9 184566
13-04-2021 09:21 1929.8 1936 1929.65 1932.8 262780

1m - historical data
time open high low close volume
13-04-2021 09:15 1924 1933.7 1924 1932 247610
13-04-2021 09:16 1932.45 1932.45 1929.4 1931.3 90342
13-04-2021 09:17 1931.35 1932.45 1930.75 1931.05 59789
13-04-2021 09:18 1931.05 1931.25 1929.45 1930.1 57982
13-04-2021 09:19 1930.1 1930.9 1928.55 1930.1 64401

Thanks.
• As you can see..the volume are not matching. 9:15
Volume for 3minute 9:15 candle looks fine to me. The volume for 9:15 3minute candle will be cumulative volume of `09:15 + 09:16 + 09:17` 1minute candle. Here cumulatoive volume for 09:15 3minute candle: 247610 + 90342 + 59789 = 397741
``13-04-2021 09:15 1924 1933.7 1924 1931.05 39774113-04-2021 09:15 1924 1933.7 1924 1932 24761013-04-2021 09:16 1932.45 1932.45 1929.4 1931.3 9034213-04-2021 09:17 1931.35 1932.45 1930.75 1931.05 59789``
Can you please guide me which is the closest correct candles (1m or 3m) to use for back testing.
Is data good (close to real) for last 5 years ? or do you have any cut off date - from which data is better?
Yes, data is good for the last 5 years or more, depending on the instrument. This thread explains this as well.
• Hi Rakesh,

I think I got where I was wrong.

Candle 9:15 - includes 9:15, 9:16, 9:17.
Candle 9:18 - includes 9:18, 9:19, 9:20.
and so on...
i.e. 9:27 includes till 9:30.

Another quick question: In some instruments - on some dates - record for 9:15 are missing. Not there in 1m and in 3m also. Any specific reason? How should I handle those.

Thanks and appreciate.
• In some instruments - on some dates - record for 9:15 are missing. Not there in 1m and in 3m also.
Can you let us know about those instruments and dates? We will check on those.
• I am attaching a file with scrip and dates (I have looked at 80% of NIFTY 500).
These records are where more than 5 records missing on 3m candles. I have not included many where 2-3 records missing, as those I can manage.

The attachment will provide you a view.
Not sure if you can fix/backfill all these, (as data in 1m is missing for these records too).

But if you can, that will be great and helpful. Thanks.
• Thanks for highlighting this. We are checking on this.