What time of day do crypto liquidations happen?

On Hyperliquid, liquidations peaked in the hour starting 15:00 UTC (11:00 in New York in summer) and were rarer on weekends. Hours, days, limits.

On Hyperliquid's smaller coins, the hour starting at 15:00 UTC was the busiest for forced closures. Across 176 coins and 114 days (June 3 to September 24, 2026), the median day in that hour held $89,181 of liquidations, almost 5 times the quietest hour's $18,001. In New York that is 11:00 in the morning in summer.

That is a small sample of days, and it describes when liquidations happened, not why. These are past cases, not a forecast.

What is a liquidation, and what is a cascade?

A liquidation happens when an exchange closes a trade because the trader's money backing it ran out. A cascade is many liquidations in a short time. Our guide to how liquidation works explains the steps.

We looked at Hyperliquid perps. A perp is a futures contract that never expires. A forced long is sold by the exchange. A forced short is bought back.

We used two sets of records. The first sums liquidated dollars minute by minute for 176 coins over 114 days. The second flags five-minute bars with unusually large liquidations on 197 coins, from May 25, 2025 to September 30, 2026.

Both sets leave out BTC, ETH, SOL and HYPE. So this is about smaller coins, not Hyperliquid's biggest markets.

Hyperliquid liquidations by UTC hour: median dayMedian liquidated dollars on a day, by UTC hour, for Hyperliquid perp coins. 00:00 $25,829; 01:00 $56,594; 02:00 $42,531; 03:00 $45,632; 04:00 $26,195; 05:00 $24,317; 06:00 $25,595; 07:00 $29,687; 08:00 $25,021; 09:00 $28,837; 10:00 $30,655; 11:00 $22,998; 12:00 $49,791; 13:00 $66,461; 14:00 $74,129; 15:00 $89,181; 16:00 $54,727; 17:00 $61,215; 18:00 $35,542; 19:00 $30,783; 20:00 $26,657; 21:00 $32,267; 22:00 $47,951; 23:00 $18,001. Busiest hour 15:00 UTC (11:00 in New York in summer); quietest 23:00 UTC. 114 days, 176 coins. Past cases, not a forecast.HYPERLIQUID / LIQUIDATIONS BY UTC HOUR$89,181liquidated on the median day in the hourstarting 15:00 UTC, the busiest (11:00in New York, summer).$0$50k$100k15:00000612151823Hour starts, UTC. Each bar: median day.Busiest: 15:00 UTC, $89,181Quietest: 23:00 UTC, $18,001Source: Hyperliquid liquidation fills, 176 perpcoins, Jun 3 to Sep 24, 2026 (114 days),Comparand archive.BTC, ETH, SOL and HYPE are not in the data.comparand.xyz
Hyperliquid liquidations by UTC hour: median day

Which hours were busiest?

We tried two ways to rank hours. The second one is the steadier one.

The median is the middle result after sorting. For each UTC hour, we took each of the 114 days, added up that hour's liquidated dollars, and found the middle day. A median is not pulled up by a few extreme days.

We also counted flagged five-minute cascade bars by hour. These are bars, not separate crashes: one market-wide move can flag many coins and many bars. We did not check each flag for missing minutes, so read the share as a count of flagged bars.

New York and London times in the table are summer times (daylight saving). In winter, both are one hour earlier. Tokyo does not change its clocks.

Which hours were busiest?
Hour starts (UTC) New York (summer) London (summer) Tokyo Median day of liquidations Share of flagged cascade bars
10:00 06:00 11:00 19:00 $30,655 2.44%
11:00 07:00 12:00 20:00 $22,998 2.99%
13:00 09:00 14:00 22:00 $66,461 4.34%
14:00 10:00 15:00 23:00 $74,129 6.02%
15:00 11:00 16:00 00:00 $89,181 6.78%
16:00 12:00 17:00 01:00 $54,727 5.93%
23:00 19:00 00:00 08:00 $18,001 3.28%

The hour starting at 15:00 UTC led in both measures. That is 11:00 in New York, 16:00 in London and midnight in Tokyo. In the cascade-bar count, 14:00 and 16:00 UTC came next. In the dollar medians, 14:00 and 13:00 UTC came next.

The two sets disagree on the third place because they cover different days and coins. That is a reason to treat the neighboring hours as a rough band, not a clean ranking. A smaller bump showed up at midnight UTC, with 5.34% of flagged bars.

How sure are we about the 15:00 UTC hour?

We re-drew the 114 days at random many times and asked each time which hour had the highest median. The hour starting at 15:00 UTC came out on top in 64% of the re-draws and in the top three in 94%. The hours starting at 14:00 and 13:00 UTC were less stable.

The result also held when we removed ZEC. Without it, the same three hours led: 15:00, 14:00 and 13:00 UTC, with a 15:00 median of $69,212.

Twenty-four hours were compared, so small gaps between neighbors are likely noise. In all but one hour, at least one of the 114 days had no liquidations at all.

Why not rank hours by total dollars?

We tried, and we do not trust it. Dollar totals are dominated by a few huge minutes.

  • The 20 largest coin-minutes held 13.84% of all liquidated dollars.
  • The 20 largest day-hours held 29.72%.
  • ZEC alone held 35.86%.
  • The average day-hour total was 8.07 times its median.

When we re-drew the days, the hour with the biggest total kept changing. So we rank hours by the median day, which a few extreme minutes cannot move much.

Which days of the week were quieter?

We counted flagged cascade bars per day of the week. Each weekday has about 70 days in the span, from May 25, 2025 to September 30, 2026.

Which days of the week were quieter?
Weekday Flagged bars per day
Monday 98.35
Tuesday 91.25
Wednesday 87.52
Thursday 110.39
Friday 124.03
Saturday 64.64
Sunday 58.31

Friday had about twice as many flagged bars as Sunday. Saturday and Sunday were the two lowest days.

The dollar data gives a noisier weekday picture. It has only 16 or 17 days for each weekday, and its biggest days are one-off events. We do not rank weekdays by dollars.

This fits a separate finding: our study of weekend volatility found Binance Bitcoin moved less per hour on weekends. That study used different data, so it supports the same direction but does not prove the same cause.

Does this match when price moves most?

It lines up with an earlier study, but not exactly. In the busiest hours for price moves, Binance Bitcoin moved most in the hour starting at 14:00 UTC, one hour before our liquidation peak.

The three busiest hours by median day start at 13:00, 14:00 and 15:00 UTC. That is 9:00 to noon in New York in summer, which covers the start of the US stock market day, which we looked at in does Bitcoin drop at the US open. We did not test whether stock trading causes the liquidation peak.

A liquidation closes a trade that lost money, so it can follow a big price move. We did not test that link, and the two studies use different exchanges and coins.

How we measured this

For the dollar data, we added up each coin's forced-long and forced-short dollars per minute from Hyperliquid's own liquidation fills, then grouped them by UTC hour and weekday. The span is June 3 to September 24, 2026, 114 days, 176 coins. For the cascade data, we flagged five-minute bars where one side's liquidations were unusually large and counted them by hour and weekday on 197 coins.

Limits: 114 days is mostly northern-hemisphere summer, so New York was on daylight time. The coin list changes as new coins list. Bars from one market-wide move cluster together. We did not test what causes the hourly pattern, and different coins have traders in different time zones.

Try this in Comparand

With Comparand connected to your AI app, you can ask about Hyperliquid:

  • "Show ZEC's liquidation totals per hour for August 22, 2026." The totals tool (hl_liquidations_series) returns one coin's liquidation totals per complete UTC hour or per day. It costs $0.005 per call on a pay-as-you-go key.
  • "Find cascade events from September 1 to September 30, 2026, on the forced-long side." The search tool (hl_cascade_search) covers up to 548 archive days and returns one row per event. You can filter by coin, direction and minimum dollars liquidated. It returns no averages, and it costs $0.005 per call.

Both tools return rows. Grouping them by hour or weekday is a step you or your agent does next. You can try 3 coins free, read the data tool docs, or check pricing.

For what price did after the biggest cascades, read our study of 300 liquidation cascades. For which side gets forced out more, read do longs or shorts get liquidated more.

The short version

  • On Hyperliquid, the hour starting at 15:00 UTC (11:00 in New York in summer) was the busiest in both measures, with a median day of $89,181.
  • The quietest hour by median day was 23:00 UTC at $18,001.
  • Fridays had 124.03 flagged cascade bars per day and Sundays 58.31, so weekends were quieter.
  • We rank hours by the median day because a few huge minutes dominate the totals.
  • BTC, ETH, SOL and HYPE are not in the data, the sample is 114 days, and these are past cases, not a forecast.

Questions people ask

What time of day do crypto liquidations happen most?

On Hyperliquid's smaller coins (not BTC, ETH, SOL or HYPE), the hour starting at 15:00 UTC was the busiest in both of our measures. Its median day held $89,181 of forced closures across 114 days (June 3 to September 24, 2026), and it had 6.78% of all flagged cascade bars. That is 11:00 in New York in summer. It is a small sample of days, and it shows when, not why.

Are there more liquidations on weekends?

Fewer. Flagged cascade bars from May 25, 2025 to September 30, 2026 averaged 124.03 on Fridays and 58.31 on Sundays. Saturday averaged 64.64. Each weekday had about 70 days in the sample.

Why do total liquidation dollars give a shaky answer?

A few huge minutes dominate the totals. The 20 largest coin-minutes held 13.84% of all liquidated dollars, and one coin, ZEC, held 35.86%. Move a few of those minutes to another hour and the top hour changes, so we ranked hours by the median day instead.

Do these hours apply to Bitcoin?

We cannot say. BTC, ETH, SOL and HYPE are not in the files we used, so the results describe 176 smaller Hyperliquid coins in the dollar data and 197 in the cascade data.

Does a busy liquidation hour tell you which way price will go?

No. We counted when forced closures happened, not what price did next. These are past cases, not a forecast.

Read next