Measuring Crypto Volatility by Time of Day: Sessions, the US Open, Weekends, Daylight Saving and the Sample-Size Trap

A 24-hour market still has a timetable

The crypto market has no open or close. Even so, people mostly act during their own daytime, and some events happen at fixed times, such as stock and futures market hours and economic data releases. So it is often said that trading crowds into certain hours with big swings while other hours are quiet. The structure of Asian, European and US hours and the time daily candles roll over were covered in the guide on 24-hour crypto market flows by time zone. This guide is the next step: rather than taking those claims on faith, it covers how to measure and read time-of-day volatility with real data. It goes through what to measure, how to view the hours that overlap the US stock open, how weekends and daylight saving shake the numbers, and how large a sample you need. In short, time-of-day statistics help you decide when to be more careful, but they are not numbers that tell you when to buy or sell.

What to measure: range, move size and trading value

Three numbers are commonly used for time-of-day volatility. The first is the range of one candle; for hourly candles, something like (high − low) ÷ open measures how far price travelled during that hour as a ratio. The second is the size of the move from open to close, taken as an absolute value regardless of direction. A large range with a small move means price swung widely and came back. The third is trading value and its share, showing how much of the period's total trading fell in that hour. The three are related but not the same, so look at them separately. Remember too that all three measure how much price moved, not which way. A high-volatility hour moved a lot both up and down; it can inform stop distances or order types, but it is not grounds for a direction.

Regular stock market hours in Korean time

Time-of-day talk in crypto usually uses each region's regular stock market hours as reference points. Below, local regular hours are converted to Korean time. Korea and Japan have no daylight saving, but the UK and the US do, so those two markets shift by an hour in Korean time depending on the season. The UK and US also start and end daylight saving on different dates, so for a few weeks a year the gap between the two markets differs from usual. Holidays and shortened sessions vary by country and year, so check exchange notices. The times below are each exchange's regular session and exclude pre-market and after-hours trading. Some trading happens outside regular hours too, so moves around the open may not start exactly on the regular-session clock.

• Tokyo: opens at 9 a.m. local time and closes in the afternoon after a lunch break. Japan and Korea share the same clock, so Korean time is identical
• Seoul (Korea Exchange): 9 a.m. to 3:30 p.m. Korean time
• London: 8 a.m. to 4:30 p.m. local time. In Korean time, 4 p.m. to 0:30 a.m. the next day during UK summer time, otherwise 5 p.m. to 1:30 a.m. the next day
• New York (NYSE, Nasdaq): 9:30 a.m. to 4 p.m. local time. In Korean time, 10:30 p.m. to 5 a.m. the next day during US daylight saving, otherwise 11:30 p.m. to 6 a.m. the next day

The hours that overlap the US open

The hour mentioned most in crypto time-of-day talk is around the US stock market open. When US regular hours begin, orders from participants who also trade stocks come in, and many days have US economic releases clustered around it. When both the UK and the US are on summer time, London and New York are open together from about 10:30 p.m. to 0:30 a.m. Korean time; when neither is, from about 11:30 p.m. to 1:30 a.m. For these reasons the data sometimes shows larger trading value and range in this window, but being structurally plausible is not the same as always being true. Busy hours can bring big moves, yet they are also when order books are deeper and the same order slips less. Splitting the first hour after the open, the few hours after, and the close often shows that the same 'US hours' have different characters.

How weekends and daylight saving skew the picture

On weekends all stock markets are closed and participants thin out, so crypto trading tends to get lighter. Thin trading can leave long wicks from small orders even in normally quiet hours, so weekend cells grow or shrink in range for different reasons than weekdays. Read weekday statistics separately from weekends. Also, just as Monday early morning in Korea is Sunday afternoon in the US, which time zone you use to cut the day puts the same hour into different weekday cells. Daylight saving creates a subtler skew. The US open falls at 10:30 p.m. or 11:30 p.m. Korean time depending on the season, so if you group more than half a year into Korean-time cells, the same 'US open' event splits across two cells. The peak on a heatmap can then look smeared over two cells rather than one. Splitting the periods before and after the change, or switching to UTC, lets you separate the effect.

How to read a time-of-day heatmap

A weekday-by-hour heatmap shows a lot at a glance, but each cell holds few samples, so it exaggerates easily. Twenty-four hours by seven days makes 168 cells; over a 30-day period, each same-weekday, same-hour cell holds only four or five candles. A single big event among them is enough to make that cell's average noticeably brighter. Reading in the order below reduces the chance of being fooled by a cell that spiked by chance. Heatmap colors also usually show relative size among cells, so even the brightest cell may differ only slightly in reality. Make a habit of checking the number behind a cell rather than its color.

• First check how many samples each cell holds
• Switch the period between 30, 90 and 180 days and see whether the same shape survives
• Put range and move size side by side to tell hours that merely swung
• Check against trading-value share to see whether a cell spiked because trading was thin
• Compare with another coin to separate a coin-specific shape from a market-wide one

Common mistakes

Time-of-day statistics make plausible conclusions easy to reach, and the mistakes below come just as easily. Averages in particular are pulled hard by a few large moves, so an hour that had one or two sharp swings can look like an hour that always moves a lot. Before drawing a conclusion, think once more about what happened in that cell and whether there is any reason for it to happen at the same hour again. When you see time-of-day statistics posted by others, first check whether they state which coin, which period and which time zone they counted; if not, it is better not to rely on them until you have measured it yourself.

• Reading high-volatility hours as rising hours or falling hours
• Treating values from cells with only a few samples as rules
• Missing that a few big events pushed the average up
• Assuming stock markets were open on holidays or shortened days
• Assuming one exchange's won market has the same shape as another exchange's dollar market
• Thinking a past pattern will carry on unchanged

Using this site's Crypto Session Clock

This site's Crypto Session Clock shows Korean time and UTC at the top, marks which of the Asian, European and US sessions is active, and flags the overlap when London and New York are open together. Each market card for Tokyo, Seoul, London and New York shows open, lunch break, closed or weekend status, the time until the next open or close, whether daylight saving applies and the next clock-change date; daylight saving shifts are applied automatically from the browser's time zone data. Holidays and shortened sessions are not included. Below, it draws a weekday-by-hour heatmap and hourly bars from the past 30, 90 or 180 days of 1-hour candles for a chosen Binance spot coin, lets you pick range, move size, trading value or trading share as the metric, and shows the value and sample count when you click a cell. 'This hour now' shows how many times the same-hour average the current 1-hour candle has moved. Use the Volume Spike Scanner for coins whose volume suddenly jumps, and the Crypto Multi Chart to compare several coins at the same hours.

Time-of-day volatility checklist

Before letting time-of-day statistics inform your plans, go through the items below in order. Keep a pattern as reference only when you can answer yes to all of them; otherwise leave it as a shape that may still be chance. Even patterns you keep should be remeasured the same way every few months, and dropped without regret if the shape has faded. The check is not a procedure for throwing numbers away but for testing how solid the shape you saw really is. Most time-of-day patterns turn out fuzzier than expected once they go through it.

• You know whether the display time zone is Korean time or UTC
• You confirmed a similar shape survives when the period changes
• You checked the sample count in each cell
• You checked whether a daylight saving change falls inside the period
• You read weekends and weekdays separately
• You do not use volatility as grounds for direction

Limits and disclaimer

Time-of-day volatility statistics are only a record of which hours saw large moves in a past period. The mix of participants, traded products and exchange rules changes every year, so there is no guarantee past shapes will continue. The regular hours in this guide are general references; check holidays, unscheduled closures, shortened sessions and rule changes in official exchange notices. Data comes from the exchange's public API and can be delayed or cut off, and the values of an unfinished candle keep changing until it closes. Spreads and slippage also vary by hour, so check the order book at the time before placing a real order. This guide explains how to read time-of-day statistics; it is not investment advice and does not suggest trading at any particular time. Make real decisions yourself based on your own situation.