When requesting odd-intervals, like 13 or 78 minute bars (both of which divide evenly into a 390-minute day), what is the correct way to identify the starting time for the data compression?

For example, if you start at midnight, 78 minute bar end-of-bar timestamps for bars that contain data for the U.S. market regular session would be (in bold):
01:18
02:36
03:54
05:12
06:30
07:48
09:06
10:24
11:42
13:00
14:18
15:36
16:54
18:12
19:30

However, I need bar data aligned with the 09:30 market open whose end-of-bar times are:
10:48:00
12:06:00
13:24:00
14:42:00
16:00:00

Specifying 0930 and 1600 for the HIT filter times returns bars with the correct timestamps. e.g.,
HIT,ABBV,4680,20221213 000000,20221215 000000,,093000,160000,1,3,,s,0
20221213 10:48 166.92 167.16 165.62 166.06 417,761
20221213 12:06 166.05 166.06 164.71 164.99 394,991

HOWEVER, inspecting the 1-minute data (always referring to end-of-bar timestamps), the OPEN for ABBV today's 09:31 bar was 166.95 - not 166.92 as indicated in the 78-minute bar data.

More strikingly, the 1-min bar close at 10:48 is 166.41 - not 166.06. Note that the bar ending at 10:17 closed at 166.06 - the only 1 minute bar to close at that price before 10:48.

What's going on with this data compression?
What's IQFeed's guidance on how to do this correctly?

You've got the right idea. Intervals assume a midnight start. 78 minutes divides evenly into a 390-minute day, but not a 1,440-minute day, and not in a way that 9:30 am is the start of a new bar. I used this, which is the same as your command:

HIT,ABBV,4680,20221213 000000,,,093000,160000,1,,,s,0
LH,2022-12-13 10:48:00,167.1600,165.6200,166.9200,166.0600,1134223,417761,0,
LH,2022-12-13 12:06:00,166.0600,164.7100,166.0500,164.9850,1709943,394991,0,
LH,2022-12-13 13:24:00,165.5500,164.6900,164.9850,165.0850,2106981,253186,0,
LH,2022-12-13 14:42:00,165.9400,164.8300,165.0900,164.8600,2610699,323191,0,
LH,2022-12-13 16:00:00,165.6500,164.6700,164.8900,164.9200,3536972,644521,0,
!ENDMSG!,

I'm looking into that Open price. I don't see any ticks for 166.92 during that time frame and I'm not sure where it's coming from.
Edited by DTN_Gary_Stephen on Dec 13, 2022 at 03:23 PM

For that request, is it misleading that IQFeed timestamps the bars in the way that [appears to] use a 0930 start? If the (ending) timestamp of the 78-minute bar is 10:48, then that implies that it's the 78-minutes of data from 0930 to 1048 - yet that is not the case (today).

I had a hunch that paid off. I rescaled IQFeed 1-minute data and compared the result to the 78-minute bars returned from IQFeed. As it turns out, using the request above, IQFeed uses the 0930 filter time as a start reference. The problem is that it doesn't do that for the current day!

See the attached image that makes this comparison. Except for occasional - but significant - differences on [usually] the 1048 bar, the bars are identical prior to "today". My guess is that when I request the ABBV 78 minute bars tomorrow, they'll be different than they are today. That's not good.

ABBV 78 Minute bars returned yesterday (12/13/2022) OHLCV
12/13/2022 10:48 166.92 167.16 165.62 166.06 417761
12/13/2022 12:06 166.05 166.06 164.71 164.985 394991
12/13/2022 13:24 164.985 165.55 164.69 165.085 253186
12/13/2022 14:42 165.09 165.94 164.83 164.86 323191
12/13/2022 16:00 164.89 165.65 164.67 164.92 644521

ABBV 78 Minute bars returned today (12/14/2022) OHLCV
12/13/2022 10:48 166.95 167.5 165.76 166.41 542785
12/13/2022 12:06 166.39 166.39 164.75 165.31 445391
12/13/2022 13:24 165.29 165.37 164.69 165.365 281420
12/13/2022 14:42 165.39 165.62 164.9 165.54 275937
12/13/2022 16:00 165.5491 165.94 164.67 164.92 786909

It's only 2AM EST, but the data returned for the same request has changed (and is now correct).

Finally, another image attached highlighting that the first 1048 bar occasionally is wrong, completely missing the first 78 minutes of data and instead appears to insert data for the 1-min of premarket trading prior to 0930. The data shown were part of the response to this request:

HIT,ABBV,4680,20211003 000000,20221216 000000,,093000,160000,1,7,,s,0

Any chance that these issues will be fixed?
Edited by Roberts on Dec 16, 2022 at 08:02 AM