The Weeks the CME Clock Slips in Latvia
Four weeks a year Chicago is seven hours behind Riga instead of eight and the US open arrives at 15:30. The full 2026 calendar, the conversion table and the setting that ends the problem.
There is a particular kind of annoyance that only shows up four weeks a year. You open the platform at the usual time, the candles are already there, the first impulse of the American session has come and gone, and nothing on the screen tells you why. No announcement, no notice from the broker, no change in the contract specification. The exchange did exactly what it always does. Latvia and the United States just moved their clocks on different weekends, and for those weeks everything you had memorised is an hour out.
For roughly forty eight weeks of the year Riga is eight hours ahead of Chicago and you never have to think about it. The trouble is precisely that you never have to think about it, because by the time March comes round the conversion has become a reflex, and reflexes do not check the calendar.
Where the eight hours comes from
CME Group publishes every session time in US Central Time. That is a deliberate choice on the exchange's part and a useful one: the Chicago clock is the reference for everything listed on CME, CBOT, NYMEX and COMEX, whether the product is an equity index, a Treasury note or a barrel of crude. Chicago sits six hours behind UTC in winter (CST) and five hours behind in summer (CDT).
Latvia sits two hours ahead of UTC in winter, Eastern European Time, and three hours ahead in summer, Eastern European Summer Time. Do the arithmetic in both seasons and you get the same answer: eight hours in January (plus two against minus six) and eight hours in July (plus three against minus five). Both sides move by exactly one hour and they move in the same direction, so the distance between them does not change.
It changes only when one side has already moved and the other has not. And because the two sets of dates were written independently, that happens every year, twice, on a schedule you can work out decades in advance.
A small piece of local context that is worth having. Latvia has been on Eastern European Time since the end of the Soviet period, when the Baltic states came off Moscow time, and the seasonal changes were applied somewhat unevenly through the 1990s before accession to the European Union locked Latvia into the common European dates. Since then the Latvian clock has done exactly what Directive 2000/84/EC says it does, no more and no less. The national implementing rules are published, like everything else, in the official legal database at likumi.lv, but they add nothing to the European dates.
The two rulebooks
Two texts govern the whole business, and neither was drafted with the other in mind.
The European one is Directive 2000/84/EC on summer-time arrangements. It puts the start of summer time at one in the morning UTC on the last Sunday in March and the end at one in the morning UTC on the last Sunday in October. The directive fixes a single instant across the whole Union rather than a local hour, which is why the change happens at three in the morning in Riga and at one in the morning in Dublin, on the same tick of the clock.
The American one is 15 U.S.C. 260a, which sets daylight saving time to begin at two in the morning local time on the second Sunday of March and to end at two in the morning local time on the first Sunday of November. Local time, so the change rolls westwards across the American time zones rather than happening everywhere at once.
From those two rules the consequences follow mechanically:
- In spring the Americans go first, on the second Sunday of March, and Latvia follows on the last Sunday of March. Depending on the year that leaves a gap of two or three weeks. Three weeks whenever the first Sunday of March falls on the 1st, 2nd or 3rd; two weeks otherwise.
- In autumn Latvia goes first, on the last Sunday of October, and the Americans follow on the first Sunday of November. Those are always consecutive Sundays, so the autumn slip is always exactly one week, in every year, with no exceptions.
In both cases the effect is identical. Chicago moves to seven hours behind Riga instead of eight, and every exchange time you know happens an hour earlier on the wall.
The 2026 calendar, and how to work out any other year
RTH open in the table below means the 08:30 Central open of the equity index regular session, which is what most people mean when they say the open.
| Stretch of 2026 | Chicago | Latvia | Gap | RTH open, Riga |
|---|---|---|---|---|
| 1 January to 7 March | CST | EET | 8 hours | 16:30 |
| 8 March to 28 March | CDT | EET | 7 hours | 15:30 |
| 29 March to 24 October | CDT | EEST | 8 hours | 16:30 |
| 25 October to 31 October | CDT | EET | 7 hours | 15:30 |
| 1 November to 31 December | CST | EET | 8 hours | 16:30 |
2026 was a long spring. The Americans changed on Sunday 8 March, Latvia on Sunday 29 March, giving a full three weeks at seven hours. Autumn was the standard single week, Sunday 25 October to Sunday 1 November.
To work out any year for yourself you only need the two Sundays in March. Find the second Sunday of March and the last Sunday of March: everything between them is a slip. Then find the last Sunday of October: the week that follows it is the other slip. For the rest of this decade the spring gap runs three weeks in 2030 and two weeks in 2027, 2028 and 2029, while the autumn gap is one week throughout.
Both changes land on a Sunday while Globex is closed, and both land before the Sunday evening reopen, so no session is ever cut short or stretched. The first session that behaves differently is the Sunday evening reopen, which makes it a convenient checkpoint. If you are at the screen on a Sunday night, you get the warning before Monday costs you anything.
The conversion table for Riga
Print this one. Column three is the answer for most of the year, column four is the answer during the slip weeks.
| Event | Exchange time | Riga, normal | Riga, slip weeks |
|---|---|---|---|
| Daily maintenance halt | 16:00 CT | 00:00 | 23:00 |
| Globex reopen, Sunday and daily | 17:00 CT | 01:00 | 00:00 |
| US macro releases (payrolls, CPI) | 07:30 CT | 15:30 | 14:30 |
| Equity index RTH open | 08:30 CT | 16:30 | 15:30 |
| EIA petroleum report, Wednesday | 09:30 CT | 17:30 | 16:30 |
| FOMC statement | 13:00 CT | 21:00 | 20:00 |
| Crude oil settlement | 13:30 CT | 21:30 | 20:30 |
| US cash equity close | 15:00 CT | 23:00 | 22:00 |
| Equity index RTH close | 15:15 CT | 23:15 | 22:15 |
Two practical observations about that table from a Latvian desk. The first is that the slip weeks move the open from a time most people can protect to a time most people cannot. Finishing work and sitting down for a 16:30 open is a routine. Being available at 15:30 in the middle of a Thursday afternoon in March is a different proposition entirely, and three weeks is long enough that "I will deal with it" turns into three weeks of arriving late.
The second is about daylight, which matters more at this latitude than people from further south expect. Around the winter solstice in Riga the sun comes up around nine in the morning and is gone again before four in the afternoon, so the 16:30 open happens in full darkness for weeks on end. By the March slip the days have stretched enough that the 15:30 open lands in daylight, and then Latvia springs forward and it goes back to 16:30. If your concentration follows the light, and for a lot of people here it does, that is worth planning around rather than discovering.
The five things that actually go wrong
Losing an hour is not the problem. The problem is everything built on top of that hour that nobody remembers to revisit.
One, alerts that carry a local time. The phone alarm set for 16:20, the platform alert configured for 16:25, the calendar block titled "open". All of them still work perfectly, and all of them now fire after the fact. There is no error message because nothing has failed.
Two, anything automated. Scheduled scripts, a machine set to wake, an ATM strategy with a timed exit, a news filter that blocks entries around a release. Every one of those has a time zone somewhere in its configuration. Anything expressed in Latvian local time drifts by an hour four weeks a year. Anything expressed in Central Time never drifts at all.
Three, your own statistics. This is the expensive one because it is invisible. If your journal records entries in local time and your charts are drawn in exchange time, then four weeks of records are shifted by an hour relative to the rest. Run a "performance by hour of day" analysis over that data and you get a conclusion about when your setup works that is partly an artefact of the calendar. Nothing about the market changed. Your measuring instrument did.
Four, the paperwork. Economic calendars usually convert to your browser's time zone automatically. Broker confirmations, statements and platform logs frequently do not, and are often stamped in Eastern or Central time. Reconciling a slip week trade between a statement in one time zone and a journal in another is how small discrepancies become an afternoon of work.
Five, the overnight. The reopen moves from one in the morning to midnight and the daily halt moves from midnight to eleven at night. If you habitually leave resting orders in place before going to bed, note that the halt now arrives an hour earlier than the mental model you have been running on.
The third clock, Eastern time
There is a complication sitting underneath all of this that has nothing to do with Latvia and makes the conversion harder than it needs to be. The American financial world does not speak in one clock either. It speaks in two.
The exchange publishes in Central Time because it is a Chicago institution. Almost everything else American that you read publishes in Eastern Time, because the news business, the cash equity market and the Federal Reserve all sit on the east coast. The opening bell in New York is at 09:30 Eastern. Payrolls and CPI come out at 08:30 Eastern. The Federal Reserve releases its statement at 14:00 Eastern. Crude settles at 14:30 Eastern.
Eastern is one hour ahead of Central, always, with no seasonal wrinkle between them, so converting is trivial once you know which clock the source is using. The problem is that sources very often do not say. A headline that reads "the Fed announces at two" and a chart drawn in Central Time are describing the same event with a one hour difference between them, and in a slip week you are now stacking that ambiguity on top of a seven hour conversion instead of an eight hour one. Two separate opportunities to be wrong by an hour, in the same afternoon.
The practical rule is to distrust any American time you read without a suffix. If the source is the exchange, assume Central. If the source is a news outlet, a US economic calendar or a central bank, assume Eastern and subtract an hour before you do anything else.
Put the charts on Chicago time
All of this disappears if you stop converting. The fix is a display setting and it takes a minute.
In NinjaTrader 8, open Tools, then Options, then the General tab, and set the time zone to Central Time. From that point every chart axis, every fill time, every entry in the trade performance report and every time stamp in the log is expressed in the same units CME Group uses to publish its hours. The open is at 08:30 in February and at 08:30 in August, and the March weeks simply stop being an event.
Three things to keep straight so this does not cause new confusion:
- This is not the Windows clock. You are changing how the platform displays time, not the operating system. Email, calendar, bank and everything else stay on Latvian time, which is where they belong.
- Trading hours templates are separate. Session definitions in NinjaTrader carry their own time zone, so a correct template opens and closes the session at the right instant no matter what the display is set to. Putting the display on Central just means what you read matches what the template is doing.
- Historical data does not move. Changing the display re labels the axis, it does not shift bars. A trade you took at 16:32 Riga time last November is the same trade, now shown at 08:32.
If you would rather not convert the whole platform, the halfway house is to add Chicago as a second clock in the Windows taskbar and keep it in view. Less elegant, but it means that in the slip weeks you are reading the correct answer instead of recalling the wrong one.
The slip week routine
- Put four dates in the calendar for the coming year: the second Sunday in March, the last Sunday in March, the last Sunday in October, the first Sunday in November. Between the first pair and between the second pair, Chicago is seven hours behind Riga rather than eight.
- Open the platform on each of those Sunday evenings before the reopen and check that the session starts when you expect it to.
- Rewrite every alert and reminder that contains a local hour, ideally against Central Time rather than local.
- Check the time zone on anything scheduled or automated, including the things you set up years ago and have not looked at since.
- For the spring slip, plan the working day around a 15:30 open. Three weeks is long enough to be worth rearranging something.
- Treat the first session after each flip as a reduced size session. Not because the market is more dangerous, but because you are the variable that changed.
Futures are leveraged and losses can run past the amount deposited, which is why turning up late and chasing a move that is already an hour old is a genuinely costly habit rather than a small embarrassment. Of all the things that can go wrong at a trading desk, this is the one with the cheapest fix.
This article is general information about exchange session times and the clock change rules in force in Latvia and the United States. It is not legal, tax or investment advice, and session hours, holiday schedules and early closes are set by the exchange and can change, so confirm the current calendar on the CME Group trading hours page and with your own broker before relying on any time shown here.
Educational content, not investment advice. Trading futures involves substantial risk of loss and is not suitable for every investor. Any reference to Algomyth results is hypothetical, simulated over historical data, and shown in full on the performance page; past performance does not guarantee future results.
Take the seat: watch the engines live
Three specialist engines on Nasdaq, S&P and Gold, every signal streamed the moment it happens, with the stop and the target written before the trade. From $149 a month per market.
