Time and time zones

How KDB stores times (UTC milliseconds), how to write a local time with its offset (like +05:45), the session's time zone (SET TIME ZONE, --timezone), time zone names, and time ranges with IN '...'.

• Category: Time-series
• Keywords: time, times, UTC, Z, time zone, timezone, summer time, daylight saving, DST, America/New_York, Europe/London, tzdata, zoneinfo, KDB_ZONEINFO, SET TIME ZONE, SHOW TIME ZONE, --timezone, KDB_TIMEZONE, CURRENT_DATE, today, local time, zone offset, NPT, Nepal time, +05:45, Asia/Kathmandu, Asia/Katmandu, Asia/Kolkata, IST, GMT, daylight saving, TIMESTAMP, DATE, milliseconds, microseconds, epoch, time format, date format, ISO 8601, IN 'interval', time range, interval, trading hours, 20700000, to_char, market open, market close
• See also: SAMPLE BY (time-zone), WHERE (in-interval), WHERE (between), Partitions, SORT BY and DEDUP KEY (partition-by), Date and time functions, Operators (arithmetic), Column types

KDB keeps every time in UTC (world time, the time at longitude 0). A local time is UTC plus or minus an offset: New York in summer is -04:00, Tokyo is +09:00. Most examples here use Nepal time (NPT), which is 5 hours 45 minutes ahead of UTC and has no daylight saving time. So 11:00 in Kathmandu is 05:15 UTC.

Three rules cover almost everything:

1. When you write a local time, add its zone offset, for example +05:45: '2026-09-10T11:00:00+05:45'.
2. A time without a zone is UTC: '2026-09-10 11:00:00' is 11:00 UTC, which is 16:45 in Nepal.
3. KDB always shows times in UTC, with a Z at the end.

To make CURRENT_DATE, ::date, date_trunc and EXTRACT work on the Nepal clock, set the session's time zone: SET TIME ZONE 'Asia/Kathmandu', or start the server with --timezone Asia/Kathmandu.

Example: A trading day: 11:00 to 15:00 Nepal time is 05:15 to 09:15 UTC (sample data: sample)

SELECT min(time) AS first_minute, max(time) AS last_minute, count(*) AS minutes
FROM ochl_minutes
WHERE symbol = 'CITY' AND time >= '2026-09-10T11:00:00+05:45' AND time < '2026-09-10T15:00:00+05:45';

Result:

(1 row)

Storage

Syntax:

TIMESTAMP    milliseconds since 1970-01-01 00:00:00 UTC
DATE         days since 1970-01-01

A TIMESTAMP column holds one number: milliseconds since the start of 1970 in UTC. It does not keep the time zone you wrote. A DATE column holds only a day, with no time and no zone.

• Returns: a TIMESTAMP is shown as 2026-09-10T05:15:00.000Z, a DATE as 2026-09-10.
• NULL: a time can be NULL (no value).
• Notes:
  • Milliseconds only. More digits after the second are cut: .1234567 becomes .123. QuestDB keeps microseconds; KDB does not.
  • Partitions (the files of a time-series table) are made by the UTC date. See PARTITION BY.
  • ROUND 1d on a column cuts times to the UTC day when a row is written (create-table). ROUND 1d TIME ZONE 'Asia/Kathmandu' cuts to Nepal midnight instead (TIME ZONE).
  • Adding a number to a TIMESTAMP adds milliseconds; TIMESTAMP - TIMESTAMP gives milliseconds. Adding a number to a DATE adds days (operators).

Example: The same moment written three ways is one stored value (sample data: sample)

Setup:

CREATE TABLE time_test (time TIMESTAMP, note STRING) TIMESERIES(time) PARTITION BY DAY;
INSERT INTO time_test VALUES
  ('2026-09-10T11:00:00+05:45', 'Nepal time with +05:45'),
  ('2026-09-10T05:15:00Z',      'UTC with Z'),
  (1789017300000,               'milliseconds since 1970'),
  ('2026-09-10T05:15:00.1234567Z', 'seven digits: cut to milliseconds');

SELECT time, note FROM time_test;

Result:

(4 rows)

Example: Milliseconds between two times, and one minute later (sample data: sample)

SELECT CAST('2026-09-10T09:15:00Z' AS TIMESTAMP) - CAST('2026-09-10T05:15:00Z' AS TIMESTAMP) AS ms_open,
       (CAST('2026-09-10T09:15:00Z' AS TIMESTAMP) - CAST('2026-09-10T05:15:00Z' AS TIMESTAMP)) / 3600000 AS hours_open,
       CAST('2026-09-10T11:00:00+05:45' AS TIMESTAMP) + 60000 AS one_minute_later;

Result:

(1 row)

Example: A DATE column: days, no time (sample data: sample)

Setup:

CREATE TABLE holidays (day DATE, name STRING) TIMESERIES(day) PARTITION BY YEAR;
INSERT INTO holidays VALUES ('2026-10-20', 'Dashain'), ('2026-11-08', 'Tihar');

SELECT day, day + 1 AS next_day, CAST(day AS TIMESTAMP) AS as_timestamp, name FROM holidays;

Result:

(2 rows)

Time text

Syntax:

'YYYY-MM-DD'
'YYYY-MM-DD HH:MM[:SS[.fff]]'
'YYYY-MM-DDTHH:MM[:SS[.fff]]'
'...Z'  |  '...+HH:MM'  |  '...+HHMM'  |  '...-HH:MM'
number  (milliseconds since 1970, UTC)

KDB reads a time from text in these forms. It reads them in INSERT, UPDATE, CAST, and when text is compared with a time column in WHERE.

• NULL: NULL stays NULL.
• Notes:
  • The year has 4 digits; month, day, hour, minute and second have 2 digits. '2026-9-10' and '10/09/2026' are errors.
  • T or one space goes between the date and the time. No space before the zone: '... 11:00:00 +05:45' is an error.
  • The date must exist: '2026-02-30' and hour 24 are errors.
  • Only in IN '...' can the text stop after the year or month ('2026-09'). In >=, <, = and BETWEEN the text needs at least the day.
  • Newer KDB builds also read '+05:45:00' (with seconds), which PostgreSQL drivers send.
  • Change other formats in your program before you send them.

Example: Many spellings of a time (sample data: sample)

SELECT CAST('2026-09-10' AS TIMESTAMP) AS date_only,
       CAST('2026-09-10 11:00' AS TIMESTAMP) AS no_zone_is_utc,
       CAST('2026-09-10T11:00:00+05:45' AS TIMESTAMP) AS nepal,
       CAST('2026-09-10T11:00:00+0545' AS TIMESTAMP) AS nepal_no_colon,
       CAST('2026-09-10T11:00:00-03:00' AS TIMESTAMP) AS brazil,
       CAST(1789017300000 AS TIMESTAMP) AS from_ms;

Result:

(1 row)

Example: Text compared with a time column is read as a time (sample data: sample)

SELECT time, symbol, close
FROM ochl_day
WHERE symbol = 'CITY' AND time >= '2026-09-08' AND time < '2026-09-11';

Result:

(3 rows)

Example: Month and day need two digits (sample data: sample; shows an error)

SELECT time, close FROM ochl_day WHERE time >= '2026-9-1';

Result:

Error: line 1, column 45: cannot compare TIMESTAMP '2025-12-28T00:00:00.000Z' with STRING '2026-9-1'

Example: Day/month/year is not a time (sample data: sample; shows an error)

INSERT INTO ochl_day VALUES ('10/09/2026', 'CITY', 1, 1, 1, 1, 1);

Result:

Error: column 'time' (TIMESTAMP): '10/09/2026' is not a time like '2026-09-11 11:00:00' (UTC) or '2026-09-11T11:00:00+05:45'

Example: Seconds in the zone, as PostgreSQL drivers send it (sample data: sample)

SELECT CAST('2026-09-10T11:00:00+05:45:00' AS TIMESTAMP) AS a;

Result:

(1 row)

A local time: Nepal +05:45

Syntax:

'YYYY-MM-DDTHH:MM:SS+05:45'
time + 20700000                          -- the Nepal clock time, still marked Z
SAMPLE BY interval TIME ZONE 'Asia/Kathmandu'
to_char(time, 'YYYY-MM-DD HH24:MI', 'Asia/Kathmandu')

The examples use Nepal Standard Time, UTC + 5 hours 45 minutes, all year; any other fixed offset works the same way. 5 h 45 min is 20,700,000 milliseconds.

• Notes:
  • The most common mistake: a local time written without its offset (here +05:45). KDB reads it as UTC, and the row is 5 h 45 min wrong. Fix bad rows with UPDATE t SET time = time - 20700000 WHERE ... (check the DEDUP KEY first).
  • hour(), day(), date_trunc() and is_trading_day() work on the session's time zone, which is UTC unless you set one. time IN '...' is always UTC. For the Nepal hour in a UTC session write hour(to_timezone(time, 'Asia/Kathmandu')). For Nepal days: time IN '2026-09-10' TIME ZONE 'Asia/Kathmandu', date_trunc('day', time, 'Asia/Kathmandu') (date functions).
  • time + 20700000 shows the Nepal clock, but the result still ends with Z. Use it only for showing or for hour(), never to store.
  • For candles on the Nepal clock use SAMPLE BY ... TIME ZONE.
  • The whole trading day of the sample data (05:15 to 09:15 UTC) is inside one UTC day, so time IN '2026-09-10' finds exactly one trading day.
  • to_char(t, format, 'Asia/Kathmandu') makes Nepal-time text; it is in newer KDB builds (date and time functions).

Example: Trades between 11:00 and 11:05 Nepal time (sample data: sample)

SELECT time, symbol, close
FROM ochl_minutes
WHERE symbol = 'CITY'
  AND time >= '2026-09-10T11:00:00+05:45'
  AND time <  '2026-09-10T11:05:00+05:45';

Result:

(5 rows)

Example: The mistake: Nepal time without +05:45 finds nothing (sample data: sample)

SELECT count(*) AS rows_found
FROM ochl_minutes
WHERE symbol = 'CITY' AND time >= '2026-09-10 11:00:00' AND time < '2026-09-10 11:05:00';

Result:

(1 row)

Example: Volume per Nepal clock hour (sample data: sample)

SELECT hour(time + 20700000) AS npt_hour, sum(volume) AS volume
FROM ochl_minutes
WHERE symbol = 'CITY' AND time IN '2026-09-10'
GROUP BY hour(time + 20700000)
ORDER BY npt_hour;

Result:

(4 rows)

Example: Show a time as Nepal time text (sample data: sample)

SELECT timestamp, to_char(timestamp, 'YYYY-MM-DD HH24:MI', 'Asia/Kathmandu') AS nepal_time
FROM live_data
WHERE symbol = 'CITY';

Result:

(3 rows)

Example: Fix rows that were written without +05:45 (sample data: sample)

Setup:

CREATE TABLE ticks (time TIMESTAMP, symbol SYMBOL, price DOUBLE) TIMESERIES(time) PARTITION BY DAY;
INSERT INTO ticks VALUES ('2026-09-10 11:00:00', 'CITY', 510), ('2026-09-10 11:01:00', 'CITY', 511);

SELECT time, price FROM ticks;
UPDATE ticks SET time = time - 20700000 WHERE time IN '2026-09-10';
SELECT time, price FROM ticks;

Result:

SELECT time, price FROM ticks

(2 rows)

UPDATE ticks SET time = time - 20700000 WHERE time IN '2026-09-10'

2 row(s) updated

SELECT time, price FROM ticks

(2 rows)

The session's time zone: SET TIME ZONE

Syntax:

SET TIME ZONE 'Asia/Kathmandu'      -- also: SET timezone = 'Asia/Kathmandu', SET TIME ZONE '+05:45'
SET TIME ZONE LOCAL                 -- back to the server's zone (also DEFAULT, RESET TIME ZONE)
SHOW TIME ZONE                      -- also SHOW timezone, current_setting('TimeZone')
kdb serve --timezone Asia/Kathmandu -- the server's zone: env KDB_TIMEZONE, kdb.conf "timezone"; else the computer's
ALTER USER asha SET timezone = 'Asia/Kathmandu' -- one user's zone: their sessions start on it
X-KDB-Time-Zone: Asia/Kathmandu     -- HTTP header: the zone for one request

Times are still stored in UTC. The time zone only decides on which day and at which clock time a stored moment falls. A session starts with its user's zone when an admin gave the user one (ALTER USER ... SET timezone), else the server's: kdb serve --timezone, else the computer's own zone (SHOW TIME ZONE says which; a container without a zone is UTC). SET TIME ZONE changes it for the rest of the session: a PostgreSQL connection, an HTTP session (X-KDB-Session), or the rest of one /exec request.

• Returns: SHOW TIME ZONE gives one row with the column TimeZone.
• NULL: not affected.
• Notes:
  • What stays in UTC: how times are stored and shown in results (...Z), now(), time text without an offset ('2026-09-10 11:00' is still 11:00 UTC), time IN '...', SAMPLE BY (add TIME ZONE to it), partitions, and the buckets of a materialized view.
  • A zone written in a call wins: date_trunc('day', t, 'UTC') and to_char(t, format, 'UTC') keep UTC.
  • A value that is already a clock time is not moved again: now() AT TIME ZONE 'Asia/Kathmandu', to_timezone(...) and LOCALTIMESTAMP.
  • Zones with summer time ('America/New_York', 'Europe/London') work too: each moment is shown on the offset in force at that moment (summer time).
  • The PostgreSQL port tells the driver the zone (TimeZone) when it connects and after each SET TIME ZONE. With --pg-timestamptz, TIMESTAMP columns are sent as text on the session's clock, like 2026-09-10 11:00:00+05:45.
  • --timezone also sets the clock of timer and backup schedules: daily 15:10 runs at 15:10 on that clock. Without it, schedules use the machine's clock (UTC in a container).
  • An +05:45 offset means 5 h 45 min ahead of UTC, as everywhere in KDB. (PostgreSQL reads SET TIME ZONE '+05:45' the other way round.)

Example: Logins per UTC day and per Nepal day: 18:30 UTC is 00:15 the next day in Nepal (sample data: sample)

Setup:

CREATE TABLE logins (time TIMESTAMP, user_name STRING) TIMESERIES(time) PARTITION BY DAY;
INSERT INTO logins VALUES ('2026-09-09T18:00:00Z', 'asha'), ('2026-09-09T18:30:00Z', 'bikash'), ('2026-09-10T03:00:00Z', 'chandra');

SELECT time::date AS utc_day, count(*) AS logins FROM logins GROUP BY 1 ORDER BY 1;
SET TIME ZONE 'Asia/Kathmandu';
SELECT time::date AS nepal_day, count(*) AS logins FROM logins GROUP BY 1 ORDER BY 1;

Result:

SELECT CAST(time AS DATE) AS utc_day, count(*) AS logins FROM logins GROUP BY 1 ORDER BY 1

(2 rows)

SET timezone = 'Asia/Kathmandu'

the time zone of this session is Asia/Kathmandu; times are still stored as UTC

SELECT CAST(time AS DATE) AS nepal_day, count(*) AS logins FROM logins GROUP BY 1 ORDER BY 1

(2 rows)

Example: Today, the clock and the text of now() on the Nepal clock (sample data: sample)

SET TIME ZONE 'Asia/Kathmandu';
SELECT CURRENT_DATE AS today_in_nepal, LOCALTIMESTAMP AS nepal_clock, now() AS utc_now,
       now()::text AS as_text, current_setting('TimeZone') AS zone;
SHOW TIME ZONE;

Result:

SET timezone = 'Asia/Kathmandu'

the time zone of this session is Asia/Kathmandu; times are still stored as UTC

SELECT current_date() AS today_in_nepal, localtimestamp() AS nepal_clock, now() AS utc_now, CAST(now

(1 row)

SHOW timezone

(1 row)

Example: A day on the Nepal clock: DATE '2026-09-10' stands for Nepal midnight, 18:15 UTC on the 9th (sample data: sample)

Setup:

CREATE TABLE logins (time TIMESTAMP, user_name STRING) TIMESERIES(time) PARTITION BY DAY;
INSERT INTO logins VALUES ('2026-09-09T18:00:00Z', 'asha'), ('2026-09-09T18:30:00Z', 'bikash'), ('2026-09-10T03:00:00Z', 'chandra');

SET TIME ZONE 'Asia/Kathmandu';
SELECT time, user_name, to_char(time, 'YYYY-MM-DD HH24:MI') AS nepal_clock
FROM logins WHERE time >= DATE '2026-09-10' AND time < DATE '2026-09-11';

Result:

SET timezone = 'Asia/Kathmandu'

the time zone of this session is Asia/Kathmandu; times are still stored as UTC

SELECT time, user_name, to_char(time, 'YYYY-MM-DD HH24:MI') AS nepal_clock FROM logins WHERE time >=

(2 rows)

Example: New York in summer and in winter: noon UTC is 08:00 in July and 07:00 in January (sample data: sample)

SET TIME ZONE 'America/New_York';
SELECT to_char(TIMESTAMP '2026-07-01 12:00:00', 'HH24:MI') AS july, to_char(TIMESTAMP '2026-01-15 12:00:00', 'HH24:MI') AS january;

Result:

SET timezone = 'America/New_York'

the time zone of this session is America/New_York; times are still stored as UTC

SELECT to_char(CAST('2026-07-01 12:00:00' AS TIMESTAMP), 'HH24:MI') AS july, to_char(CAST('2026-01-1

(1 row)

Time zone names

Syntax:

'UTC' | 'GMT' | 'Z' | 'Etc/UTC'
'Asia/Kathmandu' | 'Asia/Katmandu' | 'NPT'
'Asia/Kolkata' | 'Asia/Calcutta' | 'IST' | 'Asia/Colombo'
'Asia/Dhaka' | 'Asia/Thimphu' | 'Asia/Karachi' | 'Asia/Dubai'
'Asia/Bangkok' | 'Asia/Jakarta' | 'Asia/Shanghai' | 'Asia/Hong_Kong'
'Asia/Singapore' | 'Asia/Taipei' | 'Asia/Tokyo' | 'Asia/Seoul'
'+HH:MM' | '-HH:MM'

The zone names KDB knows, for SET TIME ZONE, kdb serve --timezone, SAMPLE BY ... TIME ZONE, time IN '...' TIME ZONE, ROUND ... TIME ZONE, date_trunc(unit, t, zone), timestamp_floor, to_timezone, to_utc and to_char.

• Notes:
  • Names do not care about case: 'asia/kathmandu' works.
  • The names above have a fixed offset and need nothing more. Every other name of the IANA time zone list ('Europe/London', 'America/New_York', 'Australia/Sydney', ...) works too, with its summer time: see summer time.
  • IST here means India Standard Time.

Example: The same hour buckets with a name and with an offset (sample data: sample)

SELECT time, count(*) AS minutes FROM ochl_minutes
WHERE symbol = 'COAST' AND time IN '2026-09-10' SAMPLE BY 1h TIME ZONE 'npt';

SELECT time, count(*) AS minutes FROM ochl_minutes
WHERE symbol = 'COAST' AND time IN '2026-09-10' SAMPLE BY 1h TIME ZONE '+05:45';

Result:

SELECT time, count(*) AS minutes FROM ochl_minutes WHERE symbol = 'COAST' AND time IN '2026-09-10' S

(4 rows)

SELECT time, count(*) AS minutes FROM ochl_minutes WHERE symbol = 'COAST' AND time IN '2026-09-10' S

(4 rows)

Zones with summer time

A zone with summer time (daylight saving time) has two offsets a year: New York is -05:00 in winter and -04:00 in summer. KDB takes the rules from the IANA time zone files that every Linux keeps (package tzdata), so each moment gets the offset that was in force at that moment, in any year.

• Where the files are: KDB looks in KDB_ZONEINFO, TZDIR, /usr/share/zoneinfo, then a zoneinfo folder next to the program. The Docker image has them. Without them a zone with summer time is an error that says where KDB looked; fixed zones always work.
• A time that happens twice (the hour after the clocks go back) is the first of the two; a time that never happens (the hour skipped when they go forward) is read with the offset from before the change, as in PostgreSQL.

Example: Day buckets in New York around the March change: 8 March is 23 hours long (sample data: sample)

CREATE TABLE ticks_ny (time TIMESTAMP, v INT) TIMESERIES(time);
INSERT INTO ticks_ny VALUES ('2026-03-08T04:30:00Z', 1), ('2026-03-08T06:00:00Z', 1), ('2026-03-09T03:30:00Z', 1), ('2026-03-09T04:30:00Z', 1);
SELECT time AS day_starts_utc, count() AS ticks FROM ticks_ny SAMPLE BY 1d TIME ZONE 'America/New_York';

Result:

CREATE TABLE ticks_ny (

table 'ticks_ny' created

INSERT INTO ticks_ny VALUES ('2026-03-08T04:30:00Z', 1), ('2026-03-08T06:00:00Z', 1), ('2026-03-09T0

4 row(s) inserted

SELECT time AS day_starts_utc, count() AS ticks FROM ticks_ny SAMPLE BY 1d TIME ZONE 'America/New_Yo

(3 rows)

Example: AT TIME ZONE: noon UTC on the New York clock, summer and winter (sample data: sample)

SELECT TIMESTAMPTZ '2026-07-01 12:00:00+00' AT TIME ZONE 'America/New_York' AS july,
       TIMESTAMPTZ '2026-01-15 12:00:00+00' AT TIME ZONE 'America/New_York' AS january;

Result:

(1 row)

Example: An offset needs two digits for the hour (sample data: sample; shows an error)

SELECT time, count(*) FROM ochl_minutes SAMPLE BY 1h TIME ZONE '+5:45';

Result:

Error: offset '+5:45' must look like '00:15', '+05:45' or '-03:00'

IN 'interval' for time ranges

Syntax:

time_column [NOT] IN 'YYYY'
time_column [NOT] IN 'YYYY-MM'
time_column [NOT] IN 'YYYY-MM-DD'
time_column [NOT] IN 'YYYY-MM-DDTHH[:MM[:SS]]'
time_column [NOT] IN '...+HH:MM'

time IN '2026-09' is true for every time inside that month. The text says how big the range is: a year, a month, a day, an hour, a minute or a second. It is the easy way to filter a time-series table by time, and KDB then reads only the partitions of that range.

• Returns: TRUE or FALSE. NOT IN '...' is true outside the range.
• NULL: a NULL time gives NULL, so the row is not kept.
• Notes:
  • Without an offset the range is in UTC. With +05:45 it is in Nepal time: the same text, one more part.
  • It works for TIMESTAMP and DATE columns. For a DATE, a day that touches the range counts.
  • Better than BETWEEN for whole days: time BETWEEN '2026-09-07' AND '2026-09-09' includes the end moment 2026-09-09T00:00:00 and nothing later on that day.
  • QuestDB's forms with ; work too: '2026-09-10;1d' stretches the end by a day, and '2026-09-10T05:15;4h;1d;5' repeats a window five days in a row. See WHERE.
  • The left side must be a time: symbol IN '2026' is an error. For a list of values write IN ('a', 'b') (WHERE).

Example: A month, a day, an hour (sample data: sample)

SELECT count(*) AS days_in_sept FROM ochl_day WHERE symbol = 'CITY' AND time IN '2026-09';
SELECT count(*) AS minutes_on_the_10th FROM ochl_minutes WHERE symbol = 'CITY' AND time IN '2026-09-10';
SELECT count(*) AS minutes_12_to_13_npt FROM ochl_minutes WHERE symbol = 'CITY' AND time IN '2026-09-10T12+05:45';

Result:

SELECT count(*) AS days_in_sept FROM ochl_day WHERE symbol = 'CITY' AND time IN '2026-09'

(1 row)

SELECT count(*) AS minutes_on_the_10th FROM ochl_minutes WHERE symbol = 'CITY' AND time IN '2026-09-

(1 row)

SELECT count(*) AS minutes_12_to_13_npt FROM ochl_minutes WHERE symbol = 'CITY' AND time IN '2026-09

(1 row)

Example: One minute on the Nepal clock (sample data: sample)

SELECT time, symbol, open, close
FROM ochl_minutes
WHERE time IN '2026-09-10T11:00+05:45';

Result:

(2 rows)

Example: BETWEEN '2026-09-09' AND '2026-09-10' misses the trading hours of the 10th (sample data: sample)

SELECT count(*) AS with_between FROM ochl_minutes
WHERE symbol = 'COAST' AND time BETWEEN '2026-09-09' AND '2026-09-10';

SELECT count(*) AS two_whole_days FROM ochl_minutes
WHERE symbol = 'COAST' AND time >= '2026-09-09' AND time < '2026-09-11';

Result:

SELECT count(*) AS with_between FROM ochl_minutes WHERE symbol = 'COAST' AND time BETWEEN '2026-09-0

(1 row)

SELECT count(*) AS two_whole_days FROM ochl_minutes WHERE symbol = 'COAST' AND time >= '2026-09-09'

(1 row)

Example: Everything except one day (sample data: sample)

SELECT count(*) AS rows_not_on_the_10th FROM ochl_minutes WHERE time NOT IN '2026-09-10';

Result:

(1 row)

Example: Not a time range (sample data: sample; shows an error)

SELECT count(*) FROM ochl_day WHERE time IN '2026-09-10 to 2026-09-11';

Result:

Error: line 1, column 42: '2026-09-10 to 2026-09-11' is not a time range like '2026-09', '2026-09-11', '2026-09-11;1d' or '2026-09-11T10:00;30m;1d;5'

Example: The left side must be a time (sample data: sample; shows an error)

SELECT count(*) FROM ochl_day WHERE symbol IN '2026';

Result:

Error: line 1, column 44: IN '...' needs a time or date on the left

Times in results

Syntax:

TIMESTAMP  ->  YYYY-MM-DDTHH:MM:SS.fffZ
DATE       ->  YYYY-MM-DD

Every time in a result is shown in UTC, with 3 digits of milliseconds and a Z. The web console, kdb sql, the HTTP API and the PostgreSQL protocol all get UTC times.

• Notes:
  • Your program changes UTC to Nepal time for the screen (for example in Go: t.In(time.FixedZone("NPT", 20700))).
  • SAMPLE BY ... TIME ZONE buckets are also shown in UTC: the Nepal hour 11:00 is 05:15:00.000Z.
  • now() is the current UTC time. It is the same value in the whole statement.
  • Over the PostgreSQL protocol a TIMESTAMP is sent as timestamp without time zone (the value is UTC) unless the server runs with --pg-timestamptz. With it, the text is on the session's clock with its offset.

Example: Live snapshots were written in Nepal time; they come back in UTC (sample data: sample)

SELECT timestamp, symbol, ltp
FROM live_data
WHERE symbol = 'TELCO';

Result:

(3 rows)

Example: now() and a DATE (sample data: sample)

SELECT now() > '2026-01-01' AS is_after_new_year, CAST('2026-09-10T20:00:00+05:45' AS DATE) AS utc_date;

Result:

(1 row)