30 days, 290 samples: Beijing time 22:00 was busiest just 50 times.
I split the 1-hour candlesticks for 10 major USDT spot pairs by UTC hour. For each asset, on each complete day, I counted the hour with the highest trading volume. That gave me 290 samples.
On average, the curve looks almost too good to be true. Trading volume at 22:00 Beijing time is 1.80 times that hour’s own 24-hour average; at 23:00, it’s 1.71 times.
The quietest hour is 7:00 in the morning, at 0.66 times the average. The two ends differ by a factor of 2.72, so on the chart there’s a single spike standing out.
At a glance, it all makes sense.
But break that spike down day by day, and it’s not so consistent. Of the 290 instances, only 50 fell at 22:00. Add the 37 at 23:00, and that’s 87 in total—a little under a third.
The other 70% happened at different times.
10:00 in the morning came out on top 26 times, and 20:00 did so 22 times. The average curve describes a tendency, not a schedule.
Many people assume the hour with the highest volume is when big money enters the market. In fact, it’s quite the opposite. At 22:00 Beijing time, the average trade was worth $205.9. Across the 10 assets, the average was $227.2—9.4% below the all-day average.
It ranked last.
The hour with the highest average trade size was 15:00 Beijing time, at $255.4. Not a huge difference.
But the direction is the opposite.
Another definition makes the distinction clearer. If we define “busiest” by number of trades instead of trading volume, the number of times 22:00 Beijing time came out on top jumps from 50 to 84, and its share rises from 17.2% to 29.0%.
The hourly rankings under the two measures have a correlation coefficient of 0.80. In any case, they point in the same direction.
The hour when the most money is traded is also the hour when the most people are trading—but people are more concentrated than money.
That’s all there is to it.
That volume bar at 22:00 is built up from the number of trades. It’s not driven by a few huge orders. The average trade size for a big-ticket asset like $BTC is naturally high; when averaged across 10 assets, 22:00 shows especially clear retail-trading characteristics.
The pattern in the timing is real; I’m not hedging on that. The 50 instances are 4.14 times the 12.08 expected by chance, and the binomial test’s p-value is so small it’s hardly worth discussing.
Split the 30 days into two halves, and the share for these two hours rises from 25.0% to 34.7%. They still ranked first in both halves.
It’s real.
But are the busiest hour and the most volatile hour the same thing? No. Out of 290 samples, the hour with the highest trading volume and the hour with the largest price range were the same only 121 times, or 41.7% of the time.
By chance, that would be 4.2%, so the two are indeed related. But on 60% of days, they don’t coincide.
That’s the point.
Trading volume by itself doesn’t tell you direction. High volume at 22:00 could mean buyers are piling in, or sellers are unloading. Trading volume can’t tell you who was the aggressor. To determine direction, you’d need to look at the taker buy/sell ratio; this chart can’t tell you that.
Volume is just volume.
What I can’t confirm is how long this pattern will last. All 10 assets fell during these 30 days, and the distribution of trading activity in a one-way down month may not look the same as in a sideways month. The data can’t separate how much of the 22:00 effect comes from the routines of Asian retail traders and how much comes from where the market was in this particular move.
What’s really worth remembering is this structure: behind a curve that looks highly certain are samples that don’t follow it on 70% of days.
Any approach that mistakes an average for a rule and uses it to time trades will eventually pay the price on individual days.
Because distributions are always more honest than averages.
$BTC is now at 80,618, $ETH 2,350, and $SOL 101.9.
All three are sitting at their 30-day lows. That 22:00 volume bar will rise on schedule again tonight.
Whether it holds up depends on the day.
Data source: Binance Spot Kline API https://api.binance.com/api/v3/klines, 1-hour intervals.
719 completed bars per asset; the still-open bar was excluded.
I split the 1-hour candlesticks for 10 major USDT spot pairs by UTC hour. For each asset, on each complete day, I counted the hour with the highest trading volume. That gave me 290 samples.
On average, the curve looks almost too good to be true. Trading volume at 22:00 Beijing time is 1.80 times that hour’s own 24-hour average; at 23:00, it’s 1.71 times.
The quietest hour is 7:00 in the morning, at 0.66 times the average. The two ends differ by a factor of 2.72, so on the chart there’s a single spike standing out.
At a glance, it all makes sense.
But break that spike down day by day, and it’s not so consistent. Of the 290 instances, only 50 fell at 22:00. Add the 37 at 23:00, and that’s 87 in total—a little under a third.
The other 70% happened at different times.
10:00 in the morning came out on top 26 times, and 20:00 did so 22 times. The average curve describes a tendency, not a schedule.
Many people assume the hour with the highest volume is when big money enters the market. In fact, it’s quite the opposite. At 22:00 Beijing time, the average trade was worth $205.9. Across the 10 assets, the average was $227.2—9.4% below the all-day average.
It ranked last.
The hour with the highest average trade size was 15:00 Beijing time, at $255.4. Not a huge difference.
But the direction is the opposite.
Another definition makes the distinction clearer. If we define “busiest” by number of trades instead of trading volume, the number of times 22:00 Beijing time came out on top jumps from 50 to 84, and its share rises from 17.2% to 29.0%.
The hourly rankings under the two measures have a correlation coefficient of 0.80. In any case, they point in the same direction.
The hour when the most money is traded is also the hour when the most people are trading—but people are more concentrated than money.
That’s all there is to it.
That volume bar at 22:00 is built up from the number of trades. It’s not driven by a few huge orders. The average trade size for a big-ticket asset like $BTC is naturally high; when averaged across 10 assets, 22:00 shows especially clear retail-trading characteristics.
The pattern in the timing is real; I’m not hedging on that. The 50 instances are 4.14 times the 12.08 expected by chance, and the binomial test’s p-value is so small it’s hardly worth discussing.
Split the 30 days into two halves, and the share for these two hours rises from 25.0% to 34.7%. They still ranked first in both halves.
It’s real.
But are the busiest hour and the most volatile hour the same thing? No. Out of 290 samples, the hour with the highest trading volume and the hour with the largest price range were the same only 121 times, or 41.7% of the time.
By chance, that would be 4.2%, so the two are indeed related. But on 60% of days, they don’t coincide.
That’s the point.
Trading volume by itself doesn’t tell you direction. High volume at 22:00 could mean buyers are piling in, or sellers are unloading. Trading volume can’t tell you who was the aggressor. To determine direction, you’d need to look at the taker buy/sell ratio; this chart can’t tell you that.
Volume is just volume.
What I can’t confirm is how long this pattern will last. All 10 assets fell during these 30 days, and the distribution of trading activity in a one-way down month may not look the same as in a sideways month. The data can’t separate how much of the 22:00 effect comes from the routines of Asian retail traders and how much comes from where the market was in this particular move.
What’s really worth remembering is this structure: behind a curve that looks highly certain are samples that don’t follow it on 70% of days.
Any approach that mistakes an average for a rule and uses it to time trades will eventually pay the price on individual days.
Because distributions are always more honest than averages.
$BTC is now at 80,618, $ETH 2,350, and $SOL 101.9.
All three are sitting at their 30-day lows. That 22:00 volume bar will rise on schedule again tonight.
Whether it holds up depends on the day.
Data source: Binance Spot Kline API https://api.binance.com/api/v3/klines, 1-hour intervals.
719 completed bars per asset; the still-open bar was excluded.