The forex market trades around the clock on weekdays, but not at weekends, and different conventions for "the trading day" cause many data bugs. Understanding forex market hours and time zones keeps your charts and calculations correct.
When is the forex market open?
Trading runs from Sunday evening to Friday evening, aligned with the New York close: it opens around 21:00-22:00 UTC on Sunday and closes around 21:00-22:00 UTC on Friday, depending on daylight saving. In between, sessions in Sydney, Tokyo, London and New York overlap.
The "New York close" trading day
Most forex platforms end each trading day at 17:00 New York time. Broker servers therefore run on New York time + 7 hours (UTC+2 in winter, UTC+3 in summer), so a daily candle runs from midnight to midnight server time. MarketAPI builds daily candles the same way, so they match what traders see.
Weekend gaps
- No prices arrive at weekends; the last quote keeps the
MARKET_CLOSEDstatus. - Monday can open far from Friday's close (a gap). Do not interpolate across it.
- For tokenized assets, settle requests made while closed at the next open price.
Practical rules for developers
- Store everything in UTC; convert only for display.
- Check the status before treating a price as current.
- Do not assume 24 hourly candles per day on Sundays, Fridays or holidays.
- Expect daylight-saving shifts twice a year, on different dates in the US and Europe.
The four trading sessions
- Sydney opens the week on Sunday evening (UTC).
- Tokyo brings yen and Asian pairs to life.
- London is the largest session; liquidity peaks here.
- New York overlaps with London in the afternoon (UTC), often the most active hours of the day.
Holidays and thin markets
Around Christmas and New Year, and on some national holidays, liquidity drops and some instruments close early. Expect fewer candles and wider spreads; treat these periods carefully in backtests.
Why do my daily candles differ from another source?
Usually because the trading day ends at a different time (midnight UTC versus 17:00 New York). Neither is wrong; just stay consistent within one dataset.
More in the OHLC guide and the API documentation.



